AI cracks Roman-era board game
By AFP
February 11, 2026
This handout picture shows a smooth, white stone dating from the Roman era in Het Romeins Museum, Heerlen that has long baffled researchers - Copyright Antiquity/AFP Handout
A smooth, white stone dating from the Roman era and unearthed in the Netherlands has long baffled researchers.
Now with the help of artificial intelligence, scientists believe they have cracked the mystery: the stone is an ancient board game and they have even guessed the rules.
The circular piece of limestone has diagonal and straight lines cut into it.
Using 3D imaging, scientists discovered some lines were deeper than others, suggesting pieces were moved along them, some more than others.
“We can see wear along the lines on the stone, exactly where you would slide a piece,” said Walter Crist, an archaeologist at Leiden University who specialises in ancient games.
Other researchers at Maastricht University then used an artificial intelligence programme that can deduce the rules to ancient games.
They trained this AI, baptised Ludii, with the rules of about 100 ancient games from the same area as the Roman stone.
The computer “produced dozens of possible rule sets. It then played the game against itself and identified a few variants that are enjoyable for humans to play,” said Dennis Soemers, from Maastricht University.
They then cross-checked the possible rules with the wear on the stone to uncover the most likely set of movements in the game.
However, Soemers also sounded a note of caution.
“If you present Ludii with a line pattern like the one on the stone, it will always find game rules. Therefore, we cannot be sure that the Romans played it in precisely that way,” he said.
The aim of the “deceptively simple but thrilling strategy game” was to hunt and trap the opponent’s pieces in as few moves as possible.
The research and the possible rules were published in the journal Antiquity.
By AFP
February 11, 2026
This handout picture shows a smooth, white stone dating from the Roman era in Het Romeins Museum, Heerlen that has long baffled researchers - Copyright Antiquity/AFP Handout
A smooth, white stone dating from the Roman era and unearthed in the Netherlands has long baffled researchers.
Now with the help of artificial intelligence, scientists believe they have cracked the mystery: the stone is an ancient board game and they have even guessed the rules.
The circular piece of limestone has diagonal and straight lines cut into it.
Using 3D imaging, scientists discovered some lines were deeper than others, suggesting pieces were moved along them, some more than others.
“We can see wear along the lines on the stone, exactly where you would slide a piece,” said Walter Crist, an archaeologist at Leiden University who specialises in ancient games.
Other researchers at Maastricht University then used an artificial intelligence programme that can deduce the rules to ancient games.
They trained this AI, baptised Ludii, with the rules of about 100 ancient games from the same area as the Roman stone.
The computer “produced dozens of possible rule sets. It then played the game against itself and identified a few variants that are enjoyable for humans to play,” said Dennis Soemers, from Maastricht University.
They then cross-checked the possible rules with the wear on the stone to uncover the most likely set of movements in the game.
However, Soemers also sounded a note of caution.
“If you present Ludii with a line pattern like the one on the stone, it will always find game rules. Therefore, we cannot be sure that the Romans played it in precisely that way,” he said.
The aim of the “deceptively simple but thrilling strategy game” was to hunt and trap the opponent’s pieces in as few moves as possible.
The research and the possible rules were published in the journal Antiquity.
All-in on AI: what TikTok creator ByteDance did next
By AFP
February 13, 2026
ByteDance has the biggest AI team in Chinese tech and plans to spend billions of dollars building AI infrastructure this year - Copyright AFP/File Pedro PARDO
Luna LIN
After soaring to global attention with its hugely popular TikTok app, Chinese tech giant ByteDance is now positioning itself as a major player in the fast-evolving AI arena.
While the Beijing-based company has been embroiled in a range of legal and privacy rows linked to the social media app for years, its team has been busy branching out developing new cutting-edge products.
Among them is China’s most popular artificial intelligence chatbot, Doubao, which has built up more than 100 million daily users since its inception in 2023.
That makes it one of the world’s largest processors of AI queries, alongside OpenAI and Google.
Meanwhile, the cinematic clips created by its latest video generator, Seedance 2.0, have further raised the company’s international profile.
But like TikTok, ByteDance’s AI services could face trouble in overseas markets owing to issues from data privacy to fierce competition in the sector.
Since OpenAI’s ChatGPT revealed the powers of AI on its 2022 debut, ByteDance has believed the technology “would become an even more important application than web search”, CEO Liang Rubo said last month.
“ByteDance’s shift reflects a deliberate evolution from social media toward an AI‑native model,” Charlie Dai, vice-president and principal analyst at Forrester, told AFP.
Regulatory and political pressure on ByteDance’s enormously popular video-sharing app TikTok has fuelled the pivot, he said.
This month, the European Commission said TikTok’s “addictive features” breached online content rules, and told it to change its design or face a fine amounting to up to six percent of ByteDance’s annual global revenue.
– ‘Evolving circumstances’ –
The United States had threatened TikTok with a total ban over concerns the platform could be used to harvest Americans’ data or spread propaganda.
After lengthy top-level talks over a TikTok divestiture deal, a majority-American-owned joint venture was established in January to operate the app’s US business, with ByteDance retaining a stake of less than 20 percent.
Rocky Lee, who uses TikTok and other sites to sell Chinese digital gadgets and pet products to buyers overseas, was relieved by the US deal.
“I can now tell other traders that ‘you can go ahead and don’t have to worry about it anymore’,” Lee, who runs a chat group for cross-border sellers, told AFP.
Lee uses Doubao and other AI tools for various tasks including product selection, market research and sales script-writing.
“We used to have more than a dozen people in our team. Now I reckon maybe four to five people are sufficient,” the veteran seller from Xi’an said.
ByteDance was US chip titan Nvidia’s largest Chinese client in 2024, and it plans to spend billions of dollars on purchasing AI microchips and building AI infrastructure in 2026.
Though less prominent internationally than domestic competitors such as DeepSeek and Qwen, Doubao models process more than 50 trillion tokens, or units of text, daily.
Google said in October that it handles more than 1.3 quadrillion tokens monthly, which is roughly 43 trillion daily.
ByteDance’s focus on AI is “a well-considered decision in response to the evolving circumstances”, said Chen Yan, an AI industry analyst at research firm QuestMobile.
“They need to seek out the next generation of productivity,” with strong growth for TikTok becoming more difficult given its already huge user base.
– Big spenders –
Shen Qiajin is founder of ideaFlow, an interactive content generation platform that is a heavy user of ByteDance AI models.
“They are taking the all-in approach with AI, and they are the most aggressive player in the market,” he told AFP.
ByteDance, which has the biggest AI team in Chinese tech, sometimes pays salaries two or three times the market average to recruit top talent, said industry headhunter Shen Wei.
“From a headhunter’s perspective, ByteDance’s advantage lies in its willingness to spend big,” he said.
Bytedance has not hidden its intention to replicate TikTok’s international success with its AI ventures.
The Doubao team is now led by Alex Zhu, who co-founded the lip-syncing app Musical.ly that later merged with TikTok.
The app is called Dola, previously Cici, overseas. Like TikTok, ByteDance’s AI services could face “concerns about data governance and geopolitical frictions”, said Forrester’s Dai.
While TikTok took over a niche, untapped market, Western AI giants “know local regulatory frameworks and user demands better”, said QuestMobile’s Chen.
Competition is also heating up at home. Tencent and Alibaba have run aggressive Lunar New Year promotions, driving their chatbots to the top of Apple’s free app chart.
Like many tech companies, ByteDance is also under pressure to make running an AI chatbot app profitable.
“The real challenge for Doubao is only coming after it has surpassed 100 million daily active users,” a Doubao staffer told Chinese tech media outlet the Late Post.
By AFP
February 13, 2026
ByteDance has the biggest AI team in Chinese tech and plans to spend billions of dollars building AI infrastructure this year - Copyright AFP/File Pedro PARDO
Luna LIN
After soaring to global attention with its hugely popular TikTok app, Chinese tech giant ByteDance is now positioning itself as a major player in the fast-evolving AI arena.
While the Beijing-based company has been embroiled in a range of legal and privacy rows linked to the social media app for years, its team has been busy branching out developing new cutting-edge products.
Among them is China’s most popular artificial intelligence chatbot, Doubao, which has built up more than 100 million daily users since its inception in 2023.
That makes it one of the world’s largest processors of AI queries, alongside OpenAI and Google.
Meanwhile, the cinematic clips created by its latest video generator, Seedance 2.0, have further raised the company’s international profile.
But like TikTok, ByteDance’s AI services could face trouble in overseas markets owing to issues from data privacy to fierce competition in the sector.
Since OpenAI’s ChatGPT revealed the powers of AI on its 2022 debut, ByteDance has believed the technology “would become an even more important application than web search”, CEO Liang Rubo said last month.
“ByteDance’s shift reflects a deliberate evolution from social media toward an AI‑native model,” Charlie Dai, vice-president and principal analyst at Forrester, told AFP.
Regulatory and political pressure on ByteDance’s enormously popular video-sharing app TikTok has fuelled the pivot, he said.
This month, the European Commission said TikTok’s “addictive features” breached online content rules, and told it to change its design or face a fine amounting to up to six percent of ByteDance’s annual global revenue.
– ‘Evolving circumstances’ –
The United States had threatened TikTok with a total ban over concerns the platform could be used to harvest Americans’ data or spread propaganda.
After lengthy top-level talks over a TikTok divestiture deal, a majority-American-owned joint venture was established in January to operate the app’s US business, with ByteDance retaining a stake of less than 20 percent.
Rocky Lee, who uses TikTok and other sites to sell Chinese digital gadgets and pet products to buyers overseas, was relieved by the US deal.
“I can now tell other traders that ‘you can go ahead and don’t have to worry about it anymore’,” Lee, who runs a chat group for cross-border sellers, told AFP.
Lee uses Doubao and other AI tools for various tasks including product selection, market research and sales script-writing.
“We used to have more than a dozen people in our team. Now I reckon maybe four to five people are sufficient,” the veteran seller from Xi’an said.
ByteDance was US chip titan Nvidia’s largest Chinese client in 2024, and it plans to spend billions of dollars on purchasing AI microchips and building AI infrastructure in 2026.
Though less prominent internationally than domestic competitors such as DeepSeek and Qwen, Doubao models process more than 50 trillion tokens, or units of text, daily.
Google said in October that it handles more than 1.3 quadrillion tokens monthly, which is roughly 43 trillion daily.
ByteDance’s focus on AI is “a well-considered decision in response to the evolving circumstances”, said Chen Yan, an AI industry analyst at research firm QuestMobile.
“They need to seek out the next generation of productivity,” with strong growth for TikTok becoming more difficult given its already huge user base.
– Big spenders –
Shen Qiajin is founder of ideaFlow, an interactive content generation platform that is a heavy user of ByteDance AI models.
“They are taking the all-in approach with AI, and they are the most aggressive player in the market,” he told AFP.
ByteDance, which has the biggest AI team in Chinese tech, sometimes pays salaries two or three times the market average to recruit top talent, said industry headhunter Shen Wei.
“From a headhunter’s perspective, ByteDance’s advantage lies in its willingness to spend big,” he said.
Bytedance has not hidden its intention to replicate TikTok’s international success with its AI ventures.
The Doubao team is now led by Alex Zhu, who co-founded the lip-syncing app Musical.ly that later merged with TikTok.
The app is called Dola, previously Cici, overseas. Like TikTok, ByteDance’s AI services could face “concerns about data governance and geopolitical frictions”, said Forrester’s Dai.
While TikTok took over a niche, untapped market, Western AI giants “know local regulatory frameworks and user demands better”, said QuestMobile’s Chen.
Competition is also heating up at home. Tencent and Alibaba have run aggressive Lunar New Year promotions, driving their chatbots to the top of Apple’s free app chart.
Like many tech companies, ByteDance is also under pressure to make running an AI chatbot app profitable.
“The real challenge for Doubao is only coming after it has surpassed 100 million daily active users,” a Doubao staffer told Chinese tech media outlet the Late Post.
AI’s bitter rivalry heads to Washington
By AFP
February 13, 2026
Dario Amodei, co-founder and CEO of Anthropic is a former staffer of OpenAI, the creator of ChatGPT - Copyright GETTY IMAGES NORTH AMERICA/AFP Yana Paskova
Alex PIGMAN
Anthropic’s major donation to a political group that competes with an OpenAI-backed organization has highlighted a bitter rift over AI regulation — a key issue heading into the US midterm elections.
With the artificial intelligence industry rapidly advancing, Democrats and Republicans alike have found themselves squeezed between a powerful tech lobby flush with cash and a broadly wary public.
Leading the charge on the industry side is Leading the Future, a pro-AI super PAC backed by OpenAI’s Greg Brockman, venture capital behemoth Andreessen Horowitz, Palantir co-founder Joe Lonsdale, and AI search company Perplexity.
Brockman, OpenAI’s longtime president, and his wife Anna are also among the largest recent donors to President Donald Trump’s political coffers, to the tune of $25 million last year.
Super PACs are political organizations in the United States that can raise and spend unlimited funds for media campaigns, but not give directly to candidates.
Leading the Future raised $125 million in the second half of 2025, according to official filings, and is co-led by Josh Vlasto — a former adviser to Fairshake, the crypto-aligned super PAC whose playbook Leading the Future is looking to repeat.
That playbook proved devastatingly effective in the 2024 election cycle, when Fairshake poured money into races against candidates skeptical of cryptocurrency.
Now spooked by the prospect of a repeat in AI, Anthropic has entered the fray.
On Thursday, the company gave $20 million to a competing super PAC, Public First Action, which supports AI guardrails — effectively setting up a direct fight against Leading the Future.
The group — whose funders can remain anonymous — plans to back 30 to 50 candidates from both parties in state and federal races during the midterm cycle.
Founded in 2021 by former AI researchers, Anthropic has grown into a world-leading AI company focused on businesses and software developers.
The company, led by CEO Dario Amodei, is disdained by some in Trump’s Washington for its outspoken focus on AI safety and its warnings about the job losses that generative AI could unleash.
The Trump administration has pushed back forcefully, championing a light regulatory touch and giving AI companies free rein to release their latest models without guardrails or pre-release vetting of their products.
Anthropic CEO Dario Amodei has made several visits to Capitol Hill
Google turns to century-long debt to build AIBy AFP
February 13, 2026
Dario Amodei, co-founder and CEO of Anthropic is a former staffer of OpenAI, the creator of ChatGPT - Copyright GETTY IMAGES NORTH AMERICA/AFP Yana Paskova
Alex PIGMAN
Anthropic’s major donation to a political group that competes with an OpenAI-backed organization has highlighted a bitter rift over AI regulation — a key issue heading into the US midterm elections.
With the artificial intelligence industry rapidly advancing, Democrats and Republicans alike have found themselves squeezed between a powerful tech lobby flush with cash and a broadly wary public.
Leading the charge on the industry side is Leading the Future, a pro-AI super PAC backed by OpenAI’s Greg Brockman, venture capital behemoth Andreessen Horowitz, Palantir co-founder Joe Lonsdale, and AI search company Perplexity.
Brockman, OpenAI’s longtime president, and his wife Anna are also among the largest recent donors to President Donald Trump’s political coffers, to the tune of $25 million last year.
Super PACs are political organizations in the United States that can raise and spend unlimited funds for media campaigns, but not give directly to candidates.
Leading the Future raised $125 million in the second half of 2025, according to official filings, and is co-led by Josh Vlasto — a former adviser to Fairshake, the crypto-aligned super PAC whose playbook Leading the Future is looking to repeat.
That playbook proved devastatingly effective in the 2024 election cycle, when Fairshake poured money into races against candidates skeptical of cryptocurrency.
Now spooked by the prospect of a repeat in AI, Anthropic has entered the fray.
On Thursday, the company gave $20 million to a competing super PAC, Public First Action, which supports AI guardrails — effectively setting up a direct fight against Leading the Future.
The group — whose funders can remain anonymous — plans to back 30 to 50 candidates from both parties in state and federal races during the midterm cycle.
Founded in 2021 by former AI researchers, Anthropic has grown into a world-leading AI company focused on businesses and software developers.
The company, led by CEO Dario Amodei, is disdained by some in Trump’s Washington for its outspoken focus on AI safety and its warnings about the job losses that generative AI could unleash.
The Trump administration has pushed back forcefully, championing a light regulatory touch and giving AI companies free rein to release their latest models without guardrails or pre-release vetting of their products.
Anthropic CEO Dario Amodei has made several visits to Capitol Hill
– Copyright GETTY IMAGES NORTH AMERICA/AFP ALEX WONG
White House AI and crypto czar David Sacks recently accused the “left-wing” company of “running a sophisticated regulatory capture strategy based on fear-mongering.”
He also accused Anthropic of retaining Democratic-aligned staffers to “lobby for the old Biden AI agenda.”
The two groups are also clashing over the Trump administration’s repeated — and so far unsuccessful — efforts to ban AI legislation at the state level.
In the absence of federal action, dozens of states have introduced hundreds of proposals to regulate the technology.
– ‘Vast resources’ –
While not as well financed as its rival, Public First Action argues it has something Leading the Future does not: the backing of public opinion.
Polls show that Americans broadly favor AI safety measures and support a more cautious approach to the technology.
“At present, there are few organized efforts to help mobilize people and politicians who understand what’s at stake in AI development,” Anthropic said in a statement.
“Instead, vast resources have flowed to political organizations that oppose these efforts.”
Amodei has also made visits to Capitol Hill to meet with Democratic Senator Elizabeth Warren to back a ban on the sending of powerful chip technology from Nvidia to China, something the Trump administration supports.
The battle is already playing out in specific races. In Florida, Leading the Future is preparing to spend millions to support Byron Donalds’ campaign for governor as Republicans in the state fight over AI legislation.
In New York, Alex Bores — a pro-AI safety congressional candidate and former Palantir employee — has already faced a barrage of attack ads from the group.
“Crazy populists…could be about to break all of this and we can’t let that happen,” Palantir co-founder Lonsdale said on CNBC in November, defending Leading the Future’s mission to fight AI safety advocates.
White House AI and crypto czar David Sacks recently accused the “left-wing” company of “running a sophisticated regulatory capture strategy based on fear-mongering.”
He also accused Anthropic of retaining Democratic-aligned staffers to “lobby for the old Biden AI agenda.”
The two groups are also clashing over the Trump administration’s repeated — and so far unsuccessful — efforts to ban AI legislation at the state level.
In the absence of federal action, dozens of states have introduced hundreds of proposals to regulate the technology.
– ‘Vast resources’ –
While not as well financed as its rival, Public First Action argues it has something Leading the Future does not: the backing of public opinion.
Polls show that Americans broadly favor AI safety measures and support a more cautious approach to the technology.
“At present, there are few organized efforts to help mobilize people and politicians who understand what’s at stake in AI development,” Anthropic said in a statement.
“Instead, vast resources have flowed to political organizations that oppose these efforts.”
Amodei has also made visits to Capitol Hill to meet with Democratic Senator Elizabeth Warren to back a ban on the sending of powerful chip technology from Nvidia to China, something the Trump administration supports.
The battle is already playing out in specific races. In Florida, Leading the Future is preparing to spend millions to support Byron Donalds’ campaign for governor as Republicans in the state fight over AI legislation.
In New York, Alex Bores — a pro-AI safety congressional candidate and former Palantir employee — has already faced a barrage of attack ads from the group.
“Crazy populists…could be about to break all of this and we can’t let that happen,” Palantir co-founder Lonsdale said on CNBC in November, defending Leading the Future’s mission to fight AI safety advocates.
Samsung starts mass production of next-gen AI memory chip
By AFP
February 12, 2026
A protoype chipset featuring Samsung Electronics' high-bandwidth memory technology was unveiled in October last year. - Copyright AFP/File Jung Yeon-je
Samsung Electronics announced Thursday it had started mass production of next-generation memory chips to power artificial intelligence, touting an “industry-leading” breakthrough.
The high-bandwidth HBM4 chips are seen as a key component needed to scale-up the vast data centres powering the explosion in artificial intelligence.
US tech giant Nvidia — the world’s most valuable company — is widely expected to be one of Samsung’s main buyers.
Samsung said it had “begun mass production of its industry-leading HBM4 and has shipped commercial products to customers”.
“This achievement marks a first in the industry, securing an early leadership position in the HBM4 market,” the South Korean company said in a statement.
A global frenzy to build AI data centres has sent orders for advanced, high‑bandwidth memory microchips soaring.
Samsung said its new chip was significantly faster than older models, exceeding industry standards for processing speed by more than 40 percent.
This would satisfy “escalating demands for higher performance”, the company said.
Samsung Electronics stock was up more than six percent in afternoon trade on South Korea’s stock exchange.
The South Korean government has pledged to become one of the world’s top three AI powers, alongside the United States and China.
Samsung and its South Korean rival SK hynix are already among the leading producers of high-performance memory chips, and the two companies had raced to start HBM4 production.
Taipei-based research firm TrendForce predicts that memory chip industry revenue will surge to a global peak of more than $840 billion in 2027.
Samsung Electronics posted record quarterly profits earlier this year, riding on massive market demand for its powerful memory chips.
The company has already earmarked billions of dollars to expand chip production facilities, pledging to continue spending in “transitioning to advanced manufacturing processes and upgrading existing production lines to meet rising demand”.
Nvidia designs hardware that powers AI computing, and has an almost insatiable demand for memory chips made by the likes of Samsung and SK hynix.
The US-based company’s almost singular role in the AI revolution has taken the world by storm since the introduction of OpenAI’s ChatGPT in late 2022.
Apple, Microsoft and Amazon have also developed chips with AI in mind, but for now are stuck trying to get their hands on Nvidia’s coveted products.
Major electronics manufacturers and industry analysts have warned that chipmakers focusing on AI sales will cause higher retail prices for consumer products across the board.
By AFP
February 12, 2026
A protoype chipset featuring Samsung Electronics' high-bandwidth memory technology was unveiled in October last year. - Copyright AFP/File Jung Yeon-je
Samsung Electronics announced Thursday it had started mass production of next-generation memory chips to power artificial intelligence, touting an “industry-leading” breakthrough.
The high-bandwidth HBM4 chips are seen as a key component needed to scale-up the vast data centres powering the explosion in artificial intelligence.
US tech giant Nvidia — the world’s most valuable company — is widely expected to be one of Samsung’s main buyers.
Samsung said it had “begun mass production of its industry-leading HBM4 and has shipped commercial products to customers”.
“This achievement marks a first in the industry, securing an early leadership position in the HBM4 market,” the South Korean company said in a statement.
A global frenzy to build AI data centres has sent orders for advanced, high‑bandwidth memory microchips soaring.
Samsung said its new chip was significantly faster than older models, exceeding industry standards for processing speed by more than 40 percent.
This would satisfy “escalating demands for higher performance”, the company said.
Samsung Electronics stock was up more than six percent in afternoon trade on South Korea’s stock exchange.
The South Korean government has pledged to become one of the world’s top three AI powers, alongside the United States and China.
Samsung and its South Korean rival SK hynix are already among the leading producers of high-performance memory chips, and the two companies had raced to start HBM4 production.
Taipei-based research firm TrendForce predicts that memory chip industry revenue will surge to a global peak of more than $840 billion in 2027.
Samsung Electronics posted record quarterly profits earlier this year, riding on massive market demand for its powerful memory chips.
The company has already earmarked billions of dollars to expand chip production facilities, pledging to continue spending in “transitioning to advanced manufacturing processes and upgrading existing production lines to meet rising demand”.
Nvidia designs hardware that powers AI computing, and has an almost insatiable demand for memory chips made by the likes of Samsung and SK hynix.
The US-based company’s almost singular role in the AI revolution has taken the world by storm since the introduction of OpenAI’s ChatGPT in late 2022.
Apple, Microsoft and Amazon have also developed chips with AI in mind, but for now are stuck trying to get their hands on Nvidia’s coveted products.
Major electronics manufacturers and industry analysts have warned that chipmakers focusing on AI sales will cause higher retail prices for consumer products across the board.
By AFP
February 10, 2026
Century-long bond issues by companies are a rarity, and especially for Alphabet which has ample online ad revenue available to pay for investments rather than resorting to debt. - Copyright AFP/File Ronan LIETAR
Google-parent Alphabet will issue bonds maturing in 100 years as it continues to invest massively in infrastructure for artificial intelligence, according to data published Tuesday by Bloomberg.
The Silicon Valley internet giant reportedly aims to raise about $20 billion overall, a chunk of it by issuing bonds that mature in February of 2126, with lenders so keen for a piece of the AI action that some $100 billion orders were placed for the debt.
Alphabet did not respond to a request for comment.
Alphabet and AI race rivals including Amazon, Meta, Microsoft are investing staggering amounts in infrastructure to power the technology, banking on it paying off.
Market reaction, though, has been mixed with some investors worried spending has gone overboard.
Century-long bond issues by companies are a rarity, and especially for Alphabet which has ample online ad revenue available to pay for investments rather than resorting to debt.
But, the rush to lead in AI has changed the game, calling for unprecedented spending on data centers, energy generation and more.
Alphabet allocated $91 billion to spending on computing infrastructure last year and has told financial analysts it expects to spend from $175 billion to $185 billion on it this year.
Alphabet has ramped up longterm debt to handle the spending surge, issuing 50-year bonds late last year.
While 100-year bonds are not new, it has been decades since US companies have resorted to them.
Companies such as Disney, Coca-Cola, FedEx, Ford, and Motorola turned to such century-long debt during the 1990s.
Latam-GPT: a Latin American AI to combat US-centric bias
By AFP
February 10, 2026
Latam-GPT is partly aimed at combating bias found in primarily US-centric AI platforms - Copyright AFP/File Ronan LIETAR
Axl HERNANDEZ
Move over ChatGPT. Chile on Tuesday launched Latam-GPT, an open-source artificial intelligence model for the region, designed to combat bias inherent in a US-centric industry.
Developed by the Chilean National Center for Artificial Intelligence (CENIA), Latam-GPT uses millions of data points collected in Latin America to showcase the continent’s cultural diversity.
“Thanks to Latam-GPT, we’re positioning the region as an active and sovereign player in the economy of the future,” President Gabriel Boric said of the initiative.
“We’re at the table — we’re not on the menu,” he added.
According to Chile’s Science Minister Aldo Valle, the program was built to combat what he called prejudices and generalizations about people and countries from the region.
Latin America, he added, “cannot simply be a passive user or recipient of artificial intelligence systems. That could result in the loss of a significant part of our traditions.”
Unlike closed generative models like ChatGPT or Google’s Gemini, Latam-GPT is an open model that can be used by programmers to customize parts of the software to suit their needs.
Contributions to the project, and data for the model’s training, were provided by Latin American universities, foundations, libraries, government entities and civil society organizations in countries including Argentina, Brazil, Chile, Colombia, Ecuador, Mexico, Peru and Uruguay.
“The models developed in other parts of the world do have data from Latin America but it represent a fairly small proportion,” CENIA director Alvaro Soto noted.
This low level of diverse input is sometimes reflected in the depictions of Latin Americans by major AI models. ChatGPT, for example, portrays a typical Chilean man as a person wearing a poncho with the Andes in the background.
– Indigenous content –
Major US tech companies dominate the global AI race, with low-cost Chinese models rapidly gaining ground and Europe lagging in third place.
Other regions of the world are also embracing the importance of developing public AI models that respect their cultural norms and safety standards.
In 2023, Singapore researchers released the open-source Southeast Asian Languages in One Network, or SEA-LION model, while in Kenya, the UlizaLLama LLM provides health services for Swahili-speaking expectant mothers.
Latam-GPT has been trained on more than eight terabytes of data, equivalent to millions of books.
It was developed for a mere $550,000, sourced primarily from the Development Bank of Latin America (CAF) and CENIA’s own resources.
A first version was developed on the Amazon Web Services cloud, but in future, Latam-GPT will be trained on a supercomputer at the University of Tarapaca in northern Chile.
For now, it is trained mainly in Spanish and Portuguese content, although its developers plan to incorporate material in Indigenous Latin American languages.
– Slang and sayings –
Latam-GPT will be available free of charge to companies and public institutions to develop applications more specific to Latin America, said Soto, the CENIA director.
He cited potential applications for hospitals “with logistical problems or issues with the use of medical resources.”
Its tiny budget means Latam-GPT has “no chance” of competing against the major AI models, Alejandro Barros, a professor in the Department of Industrial Engineering at the University of Chile, told AFP.
But it has already won over Chilean serial digital entrepreneur Roberto Musso, whose company Digevo plans to use Latam-GPT to develop customer service programs for airlines or retailers.
Musso said his clients were “very interested in having their users express themselves and receive responses in the local language.”
Latam-GPT, he said, provides the ability to recognize regional “slang, idioms, and even speech rate” and avoid biases that could arise in other AI models.
By AFP
February 10, 2026
Latam-GPT is partly aimed at combating bias found in primarily US-centric AI platforms - Copyright AFP/File Ronan LIETAR
Axl HERNANDEZ
Move over ChatGPT. Chile on Tuesday launched Latam-GPT, an open-source artificial intelligence model for the region, designed to combat bias inherent in a US-centric industry.
Developed by the Chilean National Center for Artificial Intelligence (CENIA), Latam-GPT uses millions of data points collected in Latin America to showcase the continent’s cultural diversity.
“Thanks to Latam-GPT, we’re positioning the region as an active and sovereign player in the economy of the future,” President Gabriel Boric said of the initiative.
“We’re at the table — we’re not on the menu,” he added.
According to Chile’s Science Minister Aldo Valle, the program was built to combat what he called prejudices and generalizations about people and countries from the region.
Latin America, he added, “cannot simply be a passive user or recipient of artificial intelligence systems. That could result in the loss of a significant part of our traditions.”
Unlike closed generative models like ChatGPT or Google’s Gemini, Latam-GPT is an open model that can be used by programmers to customize parts of the software to suit their needs.
Contributions to the project, and data for the model’s training, were provided by Latin American universities, foundations, libraries, government entities and civil society organizations in countries including Argentina, Brazil, Chile, Colombia, Ecuador, Mexico, Peru and Uruguay.
“The models developed in other parts of the world do have data from Latin America but it represent a fairly small proportion,” CENIA director Alvaro Soto noted.
This low level of diverse input is sometimes reflected in the depictions of Latin Americans by major AI models. ChatGPT, for example, portrays a typical Chilean man as a person wearing a poncho with the Andes in the background.
– Indigenous content –
Major US tech companies dominate the global AI race, with low-cost Chinese models rapidly gaining ground and Europe lagging in third place.
Other regions of the world are also embracing the importance of developing public AI models that respect their cultural norms and safety standards.
In 2023, Singapore researchers released the open-source Southeast Asian Languages in One Network, or SEA-LION model, while in Kenya, the UlizaLLama LLM provides health services for Swahili-speaking expectant mothers.
Latam-GPT has been trained on more than eight terabytes of data, equivalent to millions of books.
It was developed for a mere $550,000, sourced primarily from the Development Bank of Latin America (CAF) and CENIA’s own resources.
A first version was developed on the Amazon Web Services cloud, but in future, Latam-GPT will be trained on a supercomputer at the University of Tarapaca in northern Chile.
For now, it is trained mainly in Spanish and Portuguese content, although its developers plan to incorporate material in Indigenous Latin American languages.
– Slang and sayings –
Latam-GPT will be available free of charge to companies and public institutions to develop applications more specific to Latin America, said Soto, the CENIA director.
He cited potential applications for hospitals “with logistical problems or issues with the use of medical resources.”
Its tiny budget means Latam-GPT has “no chance” of competing against the major AI models, Alejandro Barros, a professor in the Department of Industrial Engineering at the University of Chile, told AFP.
But it has already won over Chilean serial digital entrepreneur Roberto Musso, whose company Digevo plans to use Latam-GPT to develop customer service programs for airlines or retailers.
Musso said his clients were “very interested in having their users express themselves and receive responses in the local language.”
Latam-GPT, he said, provides the ability to recognize regional “slang, idioms, and even speech rate” and avoid biases that could arise in other AI models.
Siemens Energy trebles profit as AI boosts power demand
By AFP
February 11, 2026
Wind turbines being built at a Siemens Energy site. The firm reported suring profits amid the AI boom - Copyright AFP FOCKE STRANGMANN
German turbine maker Siemens Energy said Wednesday that its quarterly profits had almost tripled as the firm gains from surging demand for electricity driven by the artificial intelligence boom.
The company’s gas turbines are used to generate electricity for data centres that provide computing power for AI, and have been in hot demand as US tech giants like OpenAI and Meta rapidly build more of the sites.
Net profit in the group’s fiscal first quarter, to end-December, climbed to 746 million euros ($889 million) from 252 million euros a year earlier.
Orders — an indicator of future sales — increased by a third to 17.6 billion euros.
The company’s shares rose over five percent in Frankfurt trading, putting the stock up about a quarter since the start of the year and making it the best performer to date in Germany’s blue-chip DAX index.
“Siemens Energy ticked all of the major boxes that investors were looking for with these results,” Morgan Stanley analysts wrote in a note, adding that the company’s gas turbine orders were “exceptionally strong”.
US data centre electricity consumption is projected to more than triple by 2035, according to the International Energy Agency, and already accounts for six to eight percent of US electricity use.
Asked about rising orders on an earnings call, Siemens Energy CEO Christian Bruch said he thought the first-quarter figures were not “particularly strong” and that further growth could be expected.
“Demand for gas turbines is extremely high,” he said. “We’re talking about 2029 and 2030 for delivery dates.”
Siemens Energy, spun out of the broader Siemens group in 2020, said last week that it would spend $1 billion expanding its US operations, including a new equipment plant in Mississippi as part of wider plans that would create 1,500 jobs.
Its shares have increased over tenfold since 2023, when the German government had to provide the firm with credit guarantees after quality problems at its wind-turbine unit.
By AFP
February 11, 2026
Wind turbines being built at a Siemens Energy site. The firm reported suring profits amid the AI boom - Copyright AFP FOCKE STRANGMANN
German turbine maker Siemens Energy said Wednesday that its quarterly profits had almost tripled as the firm gains from surging demand for electricity driven by the artificial intelligence boom.
The company’s gas turbines are used to generate electricity for data centres that provide computing power for AI, and have been in hot demand as US tech giants like OpenAI and Meta rapidly build more of the sites.
Net profit in the group’s fiscal first quarter, to end-December, climbed to 746 million euros ($889 million) from 252 million euros a year earlier.
Orders — an indicator of future sales — increased by a third to 17.6 billion euros.
The company’s shares rose over five percent in Frankfurt trading, putting the stock up about a quarter since the start of the year and making it the best performer to date in Germany’s blue-chip DAX index.
“Siemens Energy ticked all of the major boxes that investors were looking for with these results,” Morgan Stanley analysts wrote in a note, adding that the company’s gas turbine orders were “exceptionally strong”.
US data centre electricity consumption is projected to more than triple by 2035, according to the International Energy Agency, and already accounts for six to eight percent of US electricity use.
Asked about rising orders on an earnings call, Siemens Energy CEO Christian Bruch said he thought the first-quarter figures were not “particularly strong” and that further growth could be expected.
“Demand for gas turbines is extremely high,” he said. “We’re talking about 2029 and 2030 for delivery dates.”
Siemens Energy, spun out of the broader Siemens group in 2020, said last week that it would spend $1 billion expanding its US operations, including a new equipment plant in Mississippi as part of wider plans that would create 1,500 jobs.
Its shares have increased over tenfold since 2023, when the German government had to provide the firm with credit guarantees after quality problems at its wind-turbine unit.
What does understanding human consciousness reveal about future AI?
By Dr. Tim Sandle
SCIENCE EDITOR
By Dr. Tim Sandle
SCIENCE EDITOR
DIGITAL JOURNAL
Defining consciousness
The researchers point out that explaining how consciousness emerges is now an urgent scientific and moral priority. A clearer understanding could eventually make it possible to develop scientific methods for detecting consciousness. This breakthrough would have far-reaching consequences for AI development, prenatal policy, animal welfare, medicine, mental health care, law, and emerging technologies such as brain-computer interfaces. This can also aid in understanding what it means to be human.
The scientists warn that if we become able to create consciousness — even accidentally — it would raise immense ethical challenges and even existential risk, in relation to AI.
The Challenge of Defining Sentience
Consciousness, commonly described as awareness of both the world around us and ourselves, remains one of science’s most difficult puzzles. Despite decades of research, scientists still lack agreement on how subjective experience emerges from biological processes.
To date, scientists have identified brain regions and neural activity linked to conscious experience, but major disagreements remain. Yet there continues to be a debate as to which brain systems are truly necessary for consciousness and how they interact to produce awareness. Some researchers even question whether this approach captures the problem correctly.
The new review examines the current state of consciousness science, future directions for the field, and the possible consequences if humans succeed in fully explaining or even creating consciousness. This includes the possibility of consciousness emerging in machines or in lab-grown brain-like systems known as “brain organoids.”
Societal benefit
The researchers argue that developing evidence-based tests for consciousness could transform how awareness is identified across many contexts. These tools could help detect consciousness in patients with brain injuries or dementia and determine when awareness arises in foetuses, animals, brain organoids, or even AI systems.
For example, in medicine, this could improve care for patients who are unresponsive and assumed to be unconscious. Furthermore, understanding the biological basis of subjective experience may help researchers develop better therapies for conditions such as depression, anxiety, and schizophrenia
Warning
While this would represent a major scientific advance, the researchers caution that it would also create difficult ethical and legal questions. Determining that a system is conscious would force society to reconsider how that system should be treated.
Such insights will reshape how we see ourselves and our relationship to both artificial intelligence and the natural world. In the future, for instance, AI that gives the impression of being conscious raises many societal and ethical challenges.
To fully understand what conscious AI means, the researchers argue that scientific work should place greater emphasis on phenomenology (what consciousness feels like) alongside studies of function (what consciousness does).
To read the discussion, see Frontiers in Science and the paper titled “Consciousness science: where are we, where are we going, and what if we get there?”
An international research team has revealed the first images of the Orion Nebula captured with the James Webb Space Telescope, leaving astronomers "blown away" - Copyright AFP SERGEY BOBOK
Two scientists from the University of Sydney have performed a remarkable space science feat from Earth, the BBC reports. By using AI-driven software, the researchers have successfully corrected image blurring in NASA’s James Webb Space Telescope.
This innovation, called AMIGO, fixed distortions in the telescope’s infrared camera, restoring its ultra-sharp vision without the need for a space mission. This breakthrough restored the full precision of one of the telescope’s key instruments, achieving what would once have required a costly astronaut repair mission.
The corrective algorithms devised by the researchers ‘deblur’ the data, restoring the telescope’s full potential.
The implementation of AMIGO has led to remarkable improvements in the JWST’s imaging capabilities. With this software, the telescope has successfully captured clear images of faint celestial objects, including direct images of exoplanets and detailed observations of cosmic phenomena such as black hole jets and the surface of Jupiter’s moon Io. This demonstrates the power of combining innovative software solutions with advanced astronomical techniques.
The James Webb Space Telescope (JWST) is an advanced telescope designed to conduct infrared astronomy. It is the largest telescope in space, and is equipped with high-resolution and high-sensitivity instruments, allowing it to view objects too old, distant, or faint for the Hubble Space Telescope.
Australian science
This success builds on the JWST’s only Australian-designed component, the Aperture Masking Interferometer (AMI). This feature was created by Professor Peter Tuthill from the University of Sydney’s School of Physics and the Sydney Institute for Astronomy. The AMI allows astronomers to capture ultra-high-resolution images of stars and exoplanets.
The component works by combining light from different sections of the telescope’s main mirror, a process known as interferometry. When the JWST began its scientific operations, researchers noticed that AMI’s performance was being affected by faint electronic distortions in its infrared camera detector. These distortions caused subtle image fuzziness, reminiscent of the Hubble Space Telescope’s well-known early optical flaw that had to be corrected through astronaut spacewalks.
Making the repair
Instead of attempting a physical repair, the researchers devised a purely software-based calibration technique to fix the distortion from Earth. In other words, using artificial intelligence (AI) to steer the restoration of the telescope’s intended functionality.
Their system, called AMIGO (Aperture Masking Interferometry Generative Observations), uses advanced simulations and neural networks to replicate how the telescope’s optics and electronics function in space. By pinpointing an issue where electric charge slightly spreads to neighbouring pixels — a phenomenon called the brighter-fatter effect — the team designed algorithms that digitally corrected the images, fully restoring AMI’s performance.
‘Crystal clear’
With AMIGO in use, the James Webb Space Telescope has delivered its clearest images yet, capturing faint celestial objects in unprecedented detail. This includes direct images of a dim exoplanet and a red-brown dwarf orbiting the nearby star HD 206893, about 133 light years from Earth.
Recently, using the improved calibration, the telescope produced sharp images of a black hole jet, the fiery surface of Jupiter’s moon Io, and the dust-filled stellar winds of WR 137 — showing that JWST can now probe deeper and clearer than before.
Space tattoo
The two scientists involved, Louis Desdoigts, now a postdoctoral researcher at Leiden University in the Netherlands, and his colleague Max Charles, celebrated their achievement with tattoos of the instrument they repaired inked on their arms.
The corrective research and practical solution appear in the journal arXiv, titled “AMIGO: a Data-Driven Calibration of the JWST Interferometer.”
February 13, 2026
A dancing humanoid robot gyrates to music at a fair in Beijing. - © AFP Pedro PARDO/File
Scientists warn that rapid advances in AI and neurotechnology are outpacing our understanding of consciousness, creating serious ethical risks. New research contends that developing scientific tests for awareness could transform medicine, animal welfare, law, and AI development.
Yet identifying consciousness in machines, brain organoids, or patients could also force society to rethink responsibility, rights, and moral boundaries. Also, what does it mean to be unconscious? The question of what it means to be conscious has never been more urgent, the researchers argue, or more unsettling.
A dancing humanoid robot gyrates to music at a fair in Beijing. - © AFP Pedro PARDO/File
Scientists warn that rapid advances in AI and neurotechnology are outpacing our understanding of consciousness, creating serious ethical risks. New research contends that developing scientific tests for awareness could transform medicine, animal welfare, law, and AI development.
Yet identifying consciousness in machines, brain organoids, or patients could also force society to rethink responsibility, rights, and moral boundaries. Also, what does it mean to be unconscious? The question of what it means to be conscious has never been more urgent, the researchers argue, or more unsettling.
Will machines truly think? — Image by © Tim Sandle
Defining consciousness
The researchers point out that explaining how consciousness emerges is now an urgent scientific and moral priority. A clearer understanding could eventually make it possible to develop scientific methods for detecting consciousness. This breakthrough would have far-reaching consequences for AI development, prenatal policy, animal welfare, medicine, mental health care, law, and emerging technologies such as brain-computer interfaces. This can also aid in understanding what it means to be human.
The scientists warn that if we become able to create consciousness — even accidentally — it would raise immense ethical challenges and even existential risk, in relation to AI.
The Challenge of Defining Sentience
Consciousness, commonly described as awareness of both the world around us and ourselves, remains one of science’s most difficult puzzles. Despite decades of research, scientists still lack agreement on how subjective experience emerges from biological processes.
To date, scientists have identified brain regions and neural activity linked to conscious experience, but major disagreements remain. Yet there continues to be a debate as to which brain systems are truly necessary for consciousness and how they interact to produce awareness. Some researchers even question whether this approach captures the problem correctly.
The new review examines the current state of consciousness science, future directions for the field, and the possible consequences if humans succeed in fully explaining or even creating consciousness. This includes the possibility of consciousness emerging in machines or in lab-grown brain-like systems known as “brain organoids.”
Societal benefit
The researchers argue that developing evidence-based tests for consciousness could transform how awareness is identified across many contexts. These tools could help detect consciousness in patients with brain injuries or dementia and determine when awareness arises in foetuses, animals, brain organoids, or even AI systems.
For example, in medicine, this could improve care for patients who are unresponsive and assumed to be unconscious. Furthermore, understanding the biological basis of subjective experience may help researchers develop better therapies for conditions such as depression, anxiety, and schizophrenia
Warning
While this would represent a major scientific advance, the researchers caution that it would also create difficult ethical and legal questions. Determining that a system is conscious would force society to reconsider how that system should be treated.
Such insights will reshape how we see ourselves and our relationship to both artificial intelligence and the natural world. In the future, for instance, AI that gives the impression of being conscious raises many societal and ethical challenges.
To fully understand what conscious AI means, the researchers argue that scientific work should place greater emphasis on phenomenology (what consciousness feels like) alongside studies of function (what consciousness does).
To read the discussion, see Frontiers in Science and the paper titled “Consciousness science: where are we, where are we going, and what if we get there?”
No optical illusion: AI restores James Webb telescope
By Dr. Tim Sandle
SCIENCE EDITOR
By Dr. Tim Sandle
SCIENCE EDITOR
DIGITAL JOURNAL
February 13, 2026An international research team has revealed the first images of the Orion Nebula captured with the James Webb Space Telescope, leaving astronomers "blown away" - Copyright AFP SERGEY BOBOK
Two scientists from the University of Sydney have performed a remarkable space science feat from Earth, the BBC reports. By using AI-driven software, the researchers have successfully corrected image blurring in NASA’s James Webb Space Telescope.
This innovation, called AMIGO, fixed distortions in the telescope’s infrared camera, restoring its ultra-sharp vision without the need for a space mission. This breakthrough restored the full precision of one of the telescope’s key instruments, achieving what would once have required a costly astronaut repair mission.
The corrective algorithms devised by the researchers ‘deblur’ the data, restoring the telescope’s full potential.
The implementation of AMIGO has led to remarkable improvements in the JWST’s imaging capabilities. With this software, the telescope has successfully captured clear images of faint celestial objects, including direct images of exoplanets and detailed observations of cosmic phenomena such as black hole jets and the surface of Jupiter’s moon Io. This demonstrates the power of combining innovative software solutions with advanced astronomical techniques.
The James Webb Space Telescope (JWST) is an advanced telescope designed to conduct infrared astronomy. It is the largest telescope in space, and is equipped with high-resolution and high-sensitivity instruments, allowing it to view objects too old, distant, or faint for the Hubble Space Telescope.
Australian science
This success builds on the JWST’s only Australian-designed component, the Aperture Masking Interferometer (AMI). This feature was created by Professor Peter Tuthill from the University of Sydney’s School of Physics and the Sydney Institute for Astronomy. The AMI allows astronomers to capture ultra-high-resolution images of stars and exoplanets.
The component works by combining light from different sections of the telescope’s main mirror, a process known as interferometry. When the JWST began its scientific operations, researchers noticed that AMI’s performance was being affected by faint electronic distortions in its infrared camera detector. These distortions caused subtle image fuzziness, reminiscent of the Hubble Space Telescope’s well-known early optical flaw that had to be corrected through astronaut spacewalks.
Two stars in Wolf-Rayet 140 produce shells of dust every eight years that look like rings, as seen in this image from NASA’s James Webb Space Telescope.
Credits: NASA, ESA, CSA, STScI, JPL-Caltech
Credits: NASA, ESA, CSA, STScI, JPL-Caltech
Making the repair
Instead of attempting a physical repair, the researchers devised a purely software-based calibration technique to fix the distortion from Earth. In other words, using artificial intelligence (AI) to steer the restoration of the telescope’s intended functionality.
Their system, called AMIGO (Aperture Masking Interferometry Generative Observations), uses advanced simulations and neural networks to replicate how the telescope’s optics and electronics function in space. By pinpointing an issue where electric charge slightly spreads to neighbouring pixels — a phenomenon called the brighter-fatter effect — the team designed algorithms that digitally corrected the images, fully restoring AMI’s performance.
‘Crystal clear’
With AMIGO in use, the James Webb Space Telescope has delivered its clearest images yet, capturing faint celestial objects in unprecedented detail. This includes direct images of a dim exoplanet and a red-brown dwarf orbiting the nearby star HD 206893, about 133 light years from Earth.
Recently, using the improved calibration, the telescope produced sharp images of a black hole jet, the fiery surface of Jupiter’s moon Io, and the dust-filled stellar winds of WR 137 — showing that JWST can now probe deeper and clearer than before.
Space tattoo
The two scientists involved, Louis Desdoigts, now a postdoctoral researcher at Leiden University in the Netherlands, and his colleague Max Charles, celebrated their achievement with tattoos of the instrument they repaired inked on their arms.
The corrective research and practical solution appear in the journal arXiv, titled “AMIGO: a Data-Driven Calibration of the JWST Interferometer.”
Can generative artificial intelligence systems genuinely create original ideas?
By Dr. Tim Sandle
SCIENCE EDITOR
Creative view of the periodic table. — Image by © Tim Sandle
A new Canadian study, comparing more than 100,000 people with today’s most advanced AI systems, delivers a surprising result: generative AI can now beat the average human on certain creativity tests.
Models like GPT-4 showed strong performance on tasks designed to measure original thinking and idea generation, sometimes outperforming typical human responses.
Before people begin worrying too much about an AI takeover, there remains a clear ceiling. The most creative humans — especially the top 10% — still leave AI well behind, particularly on richer creative work like poetry and storytelling.
Scientists from the University of Montreal contend that generative AI systems have now reached a level where they can outperform the average human on certain creativity measures.
At the same time, the most creative people still show a clear and consistent advantage over even the strongest AI models.
To derive at these findings researchers evaluated several leading large language models, including ChatGPT, Claude, Gemini, and others, and compared their performance with results from more than 100,000 human participants.
Methods
To evaluate creativity fairly across humans and machines, the research team used multiple methods. The primary tool was the Divergent Association Task (DAT), a widely used psychological test that measures divergent creativity, or the ability to generate diverse and original ideas from a single prompt.
Research findings highlight a clear turning point
Some AI systems, including GPT-4, exceeded average human scores on tasks designed to measure divergent linguistic creativity.
“Our study shows that some AI systems based on large language models can now outperform average human creativity on well-defined tasks,” explains Professor Karim Jerbi in a research brief. “This result may be surprising — even unsettling — but our study also highlights an equally important observation: even the best AI systems still fall short of the levels reached by the most creative humans.”
Further analysis revealed a striking pattern. While some AI models now outperform the average person, peak creativity remains firmly human.
Moreover, when researchers examined the most creative half of participants, their average scores surpassed those of every AI model tested. The gap grew even larger among the top 10 percent of the most creative individuals.
Interpretation
The researchers then explored whether AI success on this simple word association task could extend to more complex and realistic creative activities. To test this, they compared AI systems and human participants on creative writing challenges such as composing haiku (a short three-line poetic form), writing movie plot summaries, and producing short stories.
The results followed a familiar pattern. While AI systems sometimes exceeded the performance of average humans, the most skilled human creators consistently delivered stronger and more original work.
What next for AI?
These findings raise an important question. Is AI creativity fixed, or can it be shaped? The study shows that creativity in AI can be adjusted by changing technical settings, particularly the model’s temperature. This parameter controls how predictable or adventurous the generated responses are.
At lower temperature settings, AI produces safer and more conventional outputs. At higher temperatures, responses become more varied, less predictable, and more exploratory, allowing the system to move beyond familiar ideas.
It was also demonstrated that creativity is strongly influenced by how instructions are written. For example, prompts that encourage models to think about word origins and structure using etymology lead to more unexpected associations and higher creativity scores.
These results emphasise that AI creativity depends heavily on human guidance, making interaction and prompting a central part of the creative process.
The research appears in the journal Scientific Reports, titled “Divergent creativity in humans and large language models.”
By Dr. Tim Sandle
SCIENCE EDITOR
DIGITAL JOURNAL
February 12, 2026Creative view of the periodic table. — Image by © Tim Sandle
A new Canadian study, comparing more than 100,000 people with today’s most advanced AI systems, delivers a surprising result: generative AI can now beat the average human on certain creativity tests.
Models like GPT-4 showed strong performance on tasks designed to measure original thinking and idea generation, sometimes outperforming typical human responses.
Before people begin worrying too much about an AI takeover, there remains a clear ceiling. The most creative humans — especially the top 10% — still leave AI well behind, particularly on richer creative work like poetry and storytelling.
Scientists from the University of Montreal contend that generative AI systems have now reached a level where they can outperform the average human on certain creativity measures.
At the same time, the most creative people still show a clear and consistent advantage over even the strongest AI models.
To derive at these findings researchers evaluated several leading large language models, including ChatGPT, Claude, Gemini, and others, and compared their performance with results from more than 100,000 human participants.
Methods
To evaluate creativity fairly across humans and machines, the research team used multiple methods. The primary tool was the Divergent Association Task (DAT), a widely used psychological test that measures divergent creativity, or the ability to generate diverse and original ideas from a single prompt.
Tim Sandle’s scores for the Divergent Association Task, February 2026.
Image by Tim Sandle
Research findings highlight a clear turning point
Some AI systems, including GPT-4, exceeded average human scores on tasks designed to measure divergent linguistic creativity.
“Our study shows that some AI systems based on large language models can now outperform average human creativity on well-defined tasks,” explains Professor Karim Jerbi in a research brief. “This result may be surprising — even unsettling — but our study also highlights an equally important observation: even the best AI systems still fall short of the levels reached by the most creative humans.”
Further analysis revealed a striking pattern. While some AI models now outperform the average person, peak creativity remains firmly human.
Moreover, when researchers examined the most creative half of participants, their average scores surpassed those of every AI model tested. The gap grew even larger among the top 10 percent of the most creative individuals.
Interpretation
The researchers then explored whether AI success on this simple word association task could extend to more complex and realistic creative activities. To test this, they compared AI systems and human participants on creative writing challenges such as composing haiku (a short three-line poetic form), writing movie plot summaries, and producing short stories.
The results followed a familiar pattern. While AI systems sometimes exceeded the performance of average humans, the most skilled human creators consistently delivered stronger and more original work.
What next for AI?
These findings raise an important question. Is AI creativity fixed, or can it be shaped? The study shows that creativity in AI can be adjusted by changing technical settings, particularly the model’s temperature. This parameter controls how predictable or adventurous the generated responses are.
At lower temperature settings, AI produces safer and more conventional outputs. At higher temperatures, responses become more varied, less predictable, and more exploratory, allowing the system to move beyond familiar ideas.
It was also demonstrated that creativity is strongly influenced by how instructions are written. For example, prompts that encourage models to think about word origins and structure using etymology lead to more unexpected associations and higher creativity scores.
These results emphasise that AI creativity depends heavily on human guidance, making interaction and prompting a central part of the creative process.
The research appears in the journal Scientific Reports, titled “Divergent creativity in humans and large language models.”
No comments:
Post a Comment