Monsters, asteroids, vampires: AI conspiracies flood TikTok
Washington (AFP) – From vampires and wendigos to killer asteroids, TikTok users are pumping out outlandish end-of-the-world conspiracy theories, researchers say, in yet another misinformation trend on a platform whose fate in the United States hangs in the balance.
Issued on: 18/03/2024 -
Washington (AFP) – From vampires and wendigos to killer asteroids, TikTok users are pumping out outlandish end-of-the-world conspiracy theories, researchers say, in yet another misinformation trend on a platform whose fate in the United States hangs in the balance.
Issued on: 18/03/2024 -
© GREG BAKER / AFP
In the trend reported by the nonprofit Media Matters, TikTok users seek to monetize viral videos that make unfounded claims about the US government secretly capturing or preserving mythical monsters that include -– wait for it –- King Kong.
It is the latest illustration of misinformation swirling on the platform -- a stubborn issue that has been largely absent in recent policy debates as US lawmakers mull banning the Chinese-owned app on grounds of national security.
Often accompanied by spooky background music, the videos -- many of which garner millions of views -- feature imperious AI-generated voices, sometimes mimicking celebrities.
"We are all probably going to die in the next few years. Did you hear about this?" said a voice impersonating podcaster Joe Rogan in one viral video.
"There's this asteroid that is on a collision course with Earth," the voice claims, citing information leaked by a government official who stumbled upon a folder titled "keep secret from the public."
At least one account peddling that video appeared to be deactivated after AFP reached TikTok for comment.
'Highly engaging'
Conspiracy theory videos, often posted by anonymous accounts, typically had the tell-tale signs of AI-generated images such as extra fingers and distortions, said TikTok misinformation researcher Abbie Richards.
Peddling such theories can be financially rewarding, Richards said, with TikTok's "Creativity Program" designed to pay creators for content generated on the platform.
It has spawned what she called a cottage industry of conspiracy theory videos powered by artificial intelligence tools including text-to-speech applications that are widely –- and freely –- available online.
A TikTok spokeswoman insisted that "conspiracy theories are not eligible to earn money or be recommended" in user feeds.
"Harmful misinformation is prohibited, with our safety teams removing 95 percent of it proactively before it's reported," she told AFP.
Still, tutorials on platforms such as YouTube show users how to create "viral conspiracy theory videos" and profit off TikTok's Creativity Program.
One such tutorial openly instructed users to start by making up "something outrageous" such as "scientists just got caught hiding a saber-toothed tiger."
"Financially incentivizing content that is both highly engaging and cheap to manufacture creates an environment for conspiracy theories to thrive," Richards wrote in the Media Matters report.
Threat of AI
Such concerns, driven by rapid advancements in AI, are particularly high in a year of major elections around the world.
Last week, the European Union wielded its powerful Digital Services Act (DSA) to press several platforms including TikTok on the risks of AI -– including from deepfakes -- for upcoming elections in the 27-nation bloc.
In the United States, where the app has some 170 million users -- roughly half the country's population -- lawmakers last week overwhelmingly backed a bill to ban TikTok unless Chinese parent company ByteDance divested itself within six months.
The bill, which still needs to pass the more cautious upper house of the US Congress, risks riling young voters in a key election year.
US policymakers have repeatedly expressed concerns about TikTok's alleged ties to the Chinese government, user data safety and its apparent impact on national security.
According to a report from the US Office of the Director of National Intelligence, the Chinese government is using TikTok to expand its global influence operations to promote pro-Beijing narratives and undermine American democracy, including through disinformation.
"Disinformation should be part of the debate about TikTok," Aynne Kokas, a media studies professor at the University of Virginia, told AFP.
Many experts, however, as well as young users who rely on the app as their primary source of news, oppose banning TikTok, saying it's unfair to single out the platform.
"There's lots of misinformation on TikTok, just as there is on other social media platforms. Some of that misinformation is dangerous," Jameel Jaffer, director of the Knight First Amendment Institute at Columbia University, told AFP.
"(But) investing the government with the authority to suppress misinformation -- or to ban Americans from accessing platforms that host misinformation -- is not a sensible response to this problem. Nor would it be a constitutional one," he added.
© 2024 AFP
In the trend reported by the nonprofit Media Matters, TikTok users seek to monetize viral videos that make unfounded claims about the US government secretly capturing or preserving mythical monsters that include -– wait for it –- King Kong.
It is the latest illustration of misinformation swirling on the platform -- a stubborn issue that has been largely absent in recent policy debates as US lawmakers mull banning the Chinese-owned app on grounds of national security.
Often accompanied by spooky background music, the videos -- many of which garner millions of views -- feature imperious AI-generated voices, sometimes mimicking celebrities.
"We are all probably going to die in the next few years. Did you hear about this?" said a voice impersonating podcaster Joe Rogan in one viral video.
"There's this asteroid that is on a collision course with Earth," the voice claims, citing information leaked by a government official who stumbled upon a folder titled "keep secret from the public."
At least one account peddling that video appeared to be deactivated after AFP reached TikTok for comment.
'Highly engaging'
Conspiracy theory videos, often posted by anonymous accounts, typically had the tell-tale signs of AI-generated images such as extra fingers and distortions, said TikTok misinformation researcher Abbie Richards.
Peddling such theories can be financially rewarding, Richards said, with TikTok's "Creativity Program" designed to pay creators for content generated on the platform.
It has spawned what she called a cottage industry of conspiracy theory videos powered by artificial intelligence tools including text-to-speech applications that are widely –- and freely –- available online.
A TikTok spokeswoman insisted that "conspiracy theories are not eligible to earn money or be recommended" in user feeds.
"Harmful misinformation is prohibited, with our safety teams removing 95 percent of it proactively before it's reported," she told AFP.
Still, tutorials on platforms such as YouTube show users how to create "viral conspiracy theory videos" and profit off TikTok's Creativity Program.
One such tutorial openly instructed users to start by making up "something outrageous" such as "scientists just got caught hiding a saber-toothed tiger."
"Financially incentivizing content that is both highly engaging and cheap to manufacture creates an environment for conspiracy theories to thrive," Richards wrote in the Media Matters report.
Threat of AI
Such concerns, driven by rapid advancements in AI, are particularly high in a year of major elections around the world.
Last week, the European Union wielded its powerful Digital Services Act (DSA) to press several platforms including TikTok on the risks of AI -– including from deepfakes -- for upcoming elections in the 27-nation bloc.
In the United States, where the app has some 170 million users -- roughly half the country's population -- lawmakers last week overwhelmingly backed a bill to ban TikTok unless Chinese parent company ByteDance divested itself within six months.
The bill, which still needs to pass the more cautious upper house of the US Congress, risks riling young voters in a key election year.
US policymakers have repeatedly expressed concerns about TikTok's alleged ties to the Chinese government, user data safety and its apparent impact on national security.
According to a report from the US Office of the Director of National Intelligence, the Chinese government is using TikTok to expand its global influence operations to promote pro-Beijing narratives and undermine American democracy, including through disinformation.
"Disinformation should be part of the debate about TikTok," Aynne Kokas, a media studies professor at the University of Virginia, told AFP.
Many experts, however, as well as young users who rely on the app as their primary source of news, oppose banning TikTok, saying it's unfair to single out the platform.
"There's lots of misinformation on TikTok, just as there is on other social media platforms. Some of that misinformation is dangerous," Jameel Jaffer, director of the Knight First Amendment Institute at Columbia University, told AFP.
"(But) investing the government with the authority to suppress misinformation -- or to ban Americans from accessing platforms that host misinformation -- is not a sensible response to this problem. Nor would it be a constitutional one," he added.
© 2024 AFP
The Conversation
March 14, 2024
An advertisement for TikTok is displayed at Union Station in Washington on April 3, 2023 © Stefani Reynolds / AFP
The U.S. House of Representatives voted 352-65 on March 13, 2024, to require TikTok’s parent company, China-based ByteDance, to sell the app or face a nationwide ban on TikTok. President Joe Biden said on March 8 that he would sign the legislation if it reached his desk.
The popular video social media app had 149 million users in the U.S. as of January 2024. Many of them contacted Congress to protest the possibility of a ban.
The bill’s fate in the Senate is unclear. It’s also unclear whether any resulting legislation would survive a court challenge.
On May 17, 2023, Montana Gov. Greg Gianforte signed legislation banning TikTok in the state, the first total ban by a U.S. state government. The law would impose fines of US$10,000 per day on any app store that offers TikTok and on the app-maker itself if it operates in the state. Individual users would not be subject to penalties. The law was scheduled to go into effect Jan. 1, 2024, but a federal judge blocked it pending a trial to determine whether the state overstepped its authority and whether the law violates the First Amendment.
The federal government, along with many state and foreign governments and some companies, has already banned TikTok on work-provided phones. This type of ban can be effective for protecting data related to government work.
But a full national ban of the app is another matter, which raises a number of questions: What data privacy risk does TikTok pose? What could the Chinese government do with data collected by the app? Is its content recommendation algorithm dangerous? Is it legal for a government to impose a total ban on the app? And is it even possible to ban an app?
Governments around the world have been banning TikTok on government-issued phones.
Vacuuming up data
As a cybersecurity researcher, I’ve noted that every few years, a newly popular mobile app raises issues of security, privacy and data access.
Apps collect data for several reasons. Sometimes the data is used to improve the app for users. However, most apps collect data that the companies use in part to fund their operations. This revenue typically comes from targeting users with ads based on the data they collect. The questions this use of data raises are: Does the app need all this data? What does it do with the data? And how does it protect the data from others?
So what makes TikTok different from the likes of Pokemon-GO, Facebook or even your phone itself? TikTok’s privacy policy, which few people read, is a good place to start. Overall, the company is not particularly transparent about its practices. The document is too long to list here all the data it collects, which should be a warning.
There are a few items of interest in TikTok’s privacy policy besides the information you give them when you create an account – name, age, username, password, language, email, phone number, social media account information and profile image – that are concerning. This information includes location data, data from your clipboard, contact information, website tracking, plus all data you post and messages you send through the app. The company claims that current versions of the app do not collect GPS information from U.S. users.
If most apps collect data, why are governments worried about TikTok? First, they worry about the Chinese government accessing data from TikTok’s 150 million users in the U.S. There is also a concern about the algorithms used by TikTok to show content.
As a cybersecurity researcher, I’ve noted that every few years, a newly popular mobile app raises issues of security, privacy and data access.
Apps collect data for several reasons. Sometimes the data is used to improve the app for users. However, most apps collect data that the companies use in part to fund their operations. This revenue typically comes from targeting users with ads based on the data they collect. The questions this use of data raises are: Does the app need all this data? What does it do with the data? And how does it protect the data from others?
So what makes TikTok different from the likes of Pokemon-GO, Facebook or even your phone itself? TikTok’s privacy policy, which few people read, is a good place to start. Overall, the company is not particularly transparent about its practices. The document is too long to list here all the data it collects, which should be a warning.
There are a few items of interest in TikTok’s privacy policy besides the information you give them when you create an account – name, age, username, password, language, email, phone number, social media account information and profile image – that are concerning. This information includes location data, data from your clipboard, contact information, website tracking, plus all data you post and messages you send through the app. The company claims that current versions of the app do not collect GPS information from U.S. users.
If most apps collect data, why are governments worried about TikTok? First, they worry about the Chinese government accessing data from TikTok’s 150 million users in the U.S. There is also a concern about the algorithms used by TikTok to show content.
Data in the Chinese government’s hands
If the data does end up in the hands of the Chinese government, the question is how could it use the data to its benefit. The government could share it with other companies in China to help them profit, which is no different than U.S. companies sharing marketing data. The Chinese government is known for playing the long game, and data is power, so if it is collecting data, it could take years to learn how it benefits China.
One potential threat is the Chinese government using the data to spy on people, particularly people who have access to valuable information. The Justice Department is investigating TikTok’s parent company, ByteDance, for using the app to monitor U.S. journalists. The Chinese government has an extensive history of hacking U.S. government agencies and corporations, and much of that hacking has been facilitated by social engineering – the practice of using data about people to trick them into revealing more information.
The second issue that the U.S. government has raised is algorithm bias or algorithm manipulation. TikTok and most social media apps have algorithms designed to learn a user’s interests and then try to adjust the content so the user will continue to use the app. TikTok has not shared its algorithm, so it’s not clear how the app chooses a user’s content.
The algorithm could be biased in a way that influences a population to believe certain things. There are numerous allegations that TiKTok’s algorithm is biased and can reinforce negative thoughts among younger users, and be used to affect public opinion. It could be that the algorithm’s manipulative behavior is unintentional, but there is concern that the Chinese government has been using or could use the algorithm to influence people.
TikTok’s algorithm for serving you videos has also become a source of concern.
Can a government ban an app?
The pending Montana law aims to use fines to coerce companies into enforcing its ban. It’s not clear if companies will comply, and it’s unlikely that this would deter users from finding workarounds.
Meanwhile, if the federal government comes to the conclusion that TikTok should be banned, is it even possible to ban it for all of its 149 million existing U.S. users? Any such ban would likely start with blocking the distribution of the app through Apple’s and Google’s app stores. This might keep many users off the platform, but there are other ways to download and install apps for people who are determined to use them.
A more drastic method would be to force Apple and Google to change their phones to prevent TikTok from running. While I’m not a lawyer, I think this effort would fail due to legal challenges, which include First Amendment concerns. The bottom line is that an absolute ban will be tough to enforce.
There are also questions about how effective a ban would be even if it were possible. By some estimates, the Chinese government has already collected personal information on at least 80% of the U.S. population via various means. So a ban might limit the damage going forward to some degree, but the Chinese government has already collected a significant amount of data. The Chinese government – along with anyone else with money – also has access to the large market for personal data, which fuels calls for stronger data privacy rules.
The pending Montana law aims to use fines to coerce companies into enforcing its ban. It’s not clear if companies will comply, and it’s unlikely that this would deter users from finding workarounds.
Meanwhile, if the federal government comes to the conclusion that TikTok should be banned, is it even possible to ban it for all of its 149 million existing U.S. users? Any such ban would likely start with blocking the distribution of the app through Apple’s and Google’s app stores. This might keep many users off the platform, but there are other ways to download and install apps for people who are determined to use them.
A more drastic method would be to force Apple and Google to change their phones to prevent TikTok from running. While I’m not a lawyer, I think this effort would fail due to legal challenges, which include First Amendment concerns. The bottom line is that an absolute ban will be tough to enforce.
There are also questions about how effective a ban would be even if it were possible. By some estimates, the Chinese government has already collected personal information on at least 80% of the U.S. population via various means. So a ban might limit the damage going forward to some degree, but the Chinese government has already collected a significant amount of data. The Chinese government – along with anyone else with money – also has access to the large market for personal data, which fuels calls for stronger data privacy rules.
Are you at risk?
So as an average user, should you worry? Again, it is unclear what data ByteDance is collecting and if it can harm an individual. I believe the most significant risks are to people in power, whether it is political power or within a company. Their data and information could be used to gain access to other data or potentially compromise the organizations they are associated with.
The aspect of TikTok I find most concerning is the algorithm that decides what videos users see and how it can affect vulnerable groups, particularly young people. Independent of a ban, families should have conversations about TikTok and other social media platforms and how they can be detrimental to mental health. These conversations should focus on how to determine if the app is leading you down an unhealthy path.
This is an updated version of an article originally published on March 23, 2023, and updated on May 18, 2023.
So as an average user, should you worry? Again, it is unclear what data ByteDance is collecting and if it can harm an individual. I believe the most significant risks are to people in power, whether it is political power or within a company. Their data and information could be used to gain access to other data or potentially compromise the organizations they are associated with.
The aspect of TikTok I find most concerning is the algorithm that decides what videos users see and how it can affect vulnerable groups, particularly young people. Independent of a ban, families should have conversations about TikTok and other social media platforms and how they can be detrimental to mental health. These conversations should focus on how to determine if the app is leading you down an unhealthy path.
This is an updated version of an article originally published on March 23, 2023, and updated on May 18, 2023.
TikTok creators warn of economic impact if app sees ban, call it a vital space for the marginalized
Content creator Jensen Savannah, left, and her fiance and brand manager Jorge Millares, produce a client video at El Puro Cuban Restaurant, Thursday, March 14, 2024, in Charlotte, N.C.
(AP Photo/Erik Verduzco)
By Associated Press - Saturday, March 16, 2024
Alex Pearlman shut the door on dreams of a standup comedy career almost a decade ago, pivoting from the stage to an office cubicle where he worked a customer service job.
Then he started posting random jokes and commentary about pop culture and politics on TikTok. Just over 2.5 million followers later, he quit his nine-to-five and recently booked his first nationwide tour.
Pearlman is among the many TikTok creators across the U.S. outraged over a bipartisan bill passed by the House of Representatives on Wednesday that would lead to a nationwide ban of the popular video app if its China-based owner, ByteDance, doesn’t sell its stake. The bill still needs to go through the Senate, where its prospects are unclear.
Content creators say a ban would hurt countless people and businesses that rely on TikTok for a significant portion of their income, while also arguing TikTok has become an unrivaled platform for dialogue and community.
Pearlman, who lives outside Philadelphia, said TikTok has transformed his life, allowing him to live a dream, provide for his family and spend the first three months of his newborn son’s life at home. His customer service job only offered paternity leave equivalent to three weeks off, with two weeks paid.
“I don’t take a day for granted on this app, because it’s been so shocking,” said Pearlman, 39. “In reality, TikTok has been the driver of American social media for the last four years. Something will step into that place if TikTok vanishes tomorrow. Whether or not that will be better or worse, Congress has no way of knowing.”
PHOTOS: TikTok creators warn of economic impact if app sees ban, call it a vital space for the marginalized
TikTok, which launched in 2016, has skyrocketed in popularity, growing faster than Instagram, YouTube or Facebook. The push to remove the app from Chinese authority follows concerns from lawmakers, law enforcement and intelligence officials about the insecurity of user data, potential suppression of content unfavorable to the Chinese government and the possibility that the platform could boost pro-Beijing propaganda, all of which TikTok denies.
To date, the U.S. government hasn’t provided any evidence showing TikTok shared U.S. user data with Chinese authorities.
The move comes as the pandemic saw huge growth in digital marketing as people were stuck at home consuming - and creating - content at levels not seen before.
Jensen Savannah, a 29-year-old from Charlotte, began making TikToks of her travels around the Carolinas during the pandemic. Now a full-time influencer, she has tripled her income since leaving her telecommunications sales job.
“’Social media Influencer’ is almost to be looked at as the new print and the new form of radio and TV advertising,” she said. “It’s going to bring your dollar much farther than it is in traditional marketing.”
Some creators describe it as a digital equalizer of sorts, providing a platform for people of color and other marginalized groups to get opportunities and exposure.
“I’ve always had Twitter, I’ve had Facebook, I’ve had Instagram. But TikTok was the first one where, if you want to find somebody who looks like yourself and represents you in any type of way, you can find it,” said Joshua Dairen, a Black, 30-year-old content creator in Auburn, Alabama. Dairen makes videos about his state’s ghost stories, urban legends and history.
Growing up, he loved researching everything paranormal, but he didn’t see a lot of Black representation in the field. Exposure on TikTok has led to jobs writing freelance pieces and contributing to documentaries about paranormal occurrences and unsolved mysteries. The app also gave Dairen the flexibility and confidence to open his own coffee shop, where he gets visits at least once a day from fans of his work.
He thinks banning TikTok sets “a dangerous precedent about how much power our highest levels of government can wield.”
By Associated Press - Saturday, March 16, 2024
Alex Pearlman shut the door on dreams of a standup comedy career almost a decade ago, pivoting from the stage to an office cubicle where he worked a customer service job.
Then he started posting random jokes and commentary about pop culture and politics on TikTok. Just over 2.5 million followers later, he quit his nine-to-five and recently booked his first nationwide tour.
Pearlman is among the many TikTok creators across the U.S. outraged over a bipartisan bill passed by the House of Representatives on Wednesday that would lead to a nationwide ban of the popular video app if its China-based owner, ByteDance, doesn’t sell its stake. The bill still needs to go through the Senate, where its prospects are unclear.
Content creators say a ban would hurt countless people and businesses that rely on TikTok for a significant portion of their income, while also arguing TikTok has become an unrivaled platform for dialogue and community.
Pearlman, who lives outside Philadelphia, said TikTok has transformed his life, allowing him to live a dream, provide for his family and spend the first three months of his newborn son’s life at home. His customer service job only offered paternity leave equivalent to three weeks off, with two weeks paid.
“I don’t take a day for granted on this app, because it’s been so shocking,” said Pearlman, 39. “In reality, TikTok has been the driver of American social media for the last four years. Something will step into that place if TikTok vanishes tomorrow. Whether or not that will be better or worse, Congress has no way of knowing.”
PHOTOS: TikTok creators warn of economic impact if app sees ban, call it a vital space for the marginalized
TikTok, which launched in 2016, has skyrocketed in popularity, growing faster than Instagram, YouTube or Facebook. The push to remove the app from Chinese authority follows concerns from lawmakers, law enforcement and intelligence officials about the insecurity of user data, potential suppression of content unfavorable to the Chinese government and the possibility that the platform could boost pro-Beijing propaganda, all of which TikTok denies.
To date, the U.S. government hasn’t provided any evidence showing TikTok shared U.S. user data with Chinese authorities.
The move comes as the pandemic saw huge growth in digital marketing as people were stuck at home consuming - and creating - content at levels not seen before.
Jensen Savannah, a 29-year-old from Charlotte, began making TikToks of her travels around the Carolinas during the pandemic. Now a full-time influencer, she has tripled her income since leaving her telecommunications sales job.
“’Social media Influencer’ is almost to be looked at as the new print and the new form of radio and TV advertising,” she said. “It’s going to bring your dollar much farther than it is in traditional marketing.”
Some creators describe it as a digital equalizer of sorts, providing a platform for people of color and other marginalized groups to get opportunities and exposure.
“I’ve always had Twitter, I’ve had Facebook, I’ve had Instagram. But TikTok was the first one where, if you want to find somebody who looks like yourself and represents you in any type of way, you can find it,” said Joshua Dairen, a Black, 30-year-old content creator in Auburn, Alabama. Dairen makes videos about his state’s ghost stories, urban legends and history.
Growing up, he loved researching everything paranormal, but he didn’t see a lot of Black representation in the field. Exposure on TikTok has led to jobs writing freelance pieces and contributing to documentaries about paranormal occurrences and unsolved mysteries. The app also gave Dairen the flexibility and confidence to open his own coffee shop, where he gets visits at least once a day from fans of his work.
He thinks banning TikTok sets “a dangerous precedent about how much power our highest levels of government can wield.”
The state of AI infrastructure at scale: Exposing GPU utilization challenges
ByDr. Tim Sandle
March 15, 2024
Generative AI apps like ChatGPT are raising concerns about the impact of artificial intelligence on a range of issues including disinformation as well as copyright over images, sound and text
How they rank priorities when evaluating AI infrastructure solutions against their business use cases.
More specifically in relation to the compute concerns, latency was top-ranked at, followed by power consumption. To address this, the majority of executives plan to use more cloud compute and many will buy more GPU machines on-premises in 2024 (a graphics processing unit – GPU – is an electronic circuit that can perform mathematical calculations at high speed. Computing tasks like graphics rendering, machine learning, and video editing require the application of similar mathematical operations on a large dataset).
On the issue of latency, over half of respondents plan to use language models (like LLama), followed by embedding models (BERT and family) in their commercial deployments. Mitigating compute challenges will be essential in their plans.
One challenge is the global limitations in GPU supplies. A global chip shortage, triggered by the COVID-19 pandemic in 2020, severely hampered the production of GPUs. The pandemic disrupted the global supply chain, causing delays in chip production and delivery. To counter GPU scarcity, most businesses are looking for or are interested in cost-effective alternatives to GPUs.
The main challenges for operating GPUs is with job scheduling and management. This is especially with coordinating tasks and workflows within the AI/ML technology stack; something that is necessary in order to optimize GPU and compute resource allocation.
For those who already operate cloud compute systems, the main concerns are around wastage and idle costs. In addition there are misgivings about the cost of overall compute power consumption.
ByDr. Tim Sandle
March 15, 2024
Generative AI apps like ChatGPT are raising concerns about the impact of artificial intelligence on a range of issues including disinformation as well as copyright over images, sound and text
- © AFP Julio Cesar AGUILAR
The AI Infrastructure Alliance, MLOps co ClearML and chip firm FuriosaAI have teamed up as so to assess what business executives think about artificial intelligence. The output is a new report titled “The State of AI Infrastructure at Scale 2024: Unveiling Future Landscapes, Key Insights, and Business Benchmarks”. The tome includes responses from AI/ML and technology leaders across North America, Europe, and Asia Pacific, addressing issues and obstacles to scale-up.
Many executives reported that having and using Open Source technology is important for their organization. With most focused on customizing Open Source models. PyTorch is their framework of choice. PyTorch is a machine learning library used for applications such as computer vision and natural language processing.
This assessment has revealed that the biggest challenge is in scaling AI is compute limitations (an issue of both availability and cost). The next top challenge was infrastructure issues.
Central concerns are with:How executives are building their AI infrastructure.
The critical benchmarks and key challenges they face.
The AI Infrastructure Alliance, MLOps co ClearML and chip firm FuriosaAI have teamed up as so to assess what business executives think about artificial intelligence. The output is a new report titled “The State of AI Infrastructure at Scale 2024: Unveiling Future Landscapes, Key Insights, and Business Benchmarks”. The tome includes responses from AI/ML and technology leaders across North America, Europe, and Asia Pacific, addressing issues and obstacles to scale-up.
Many executives reported that having and using Open Source technology is important for their organization. With most focused on customizing Open Source models. PyTorch is their framework of choice. PyTorch is a machine learning library used for applications such as computer vision and natural language processing.
This assessment has revealed that the biggest challenge is in scaling AI is compute limitations (an issue of both availability and cost). The next top challenge was infrastructure issues.
Central concerns are with:How executives are building their AI infrastructure.
The critical benchmarks and key challenges they face.
How they rank priorities when evaluating AI infrastructure solutions against their business use cases.
More specifically in relation to the compute concerns, latency was top-ranked at, followed by power consumption. To address this, the majority of executives plan to use more cloud compute and many will buy more GPU machines on-premises in 2024 (a graphics processing unit – GPU – is an electronic circuit that can perform mathematical calculations at high speed. Computing tasks like graphics rendering, machine learning, and video editing require the application of similar mathematical operations on a large dataset).
On the issue of latency, over half of respondents plan to use language models (like LLama), followed by embedding models (BERT and family) in their commercial deployments. Mitigating compute challenges will be essential in their plans.
One challenge is the global limitations in GPU supplies. A global chip shortage, triggered by the COVID-19 pandemic in 2020, severely hampered the production of GPUs. The pandemic disrupted the global supply chain, causing delays in chip production and delivery. To counter GPU scarcity, most businesses are looking for or are interested in cost-effective alternatives to GPUs.
The main challenges for operating GPUs is with job scheduling and management. This is especially with coordinating tasks and workflows within the AI/ML technology stack; something that is necessary in order to optimize GPU and compute resource allocation.
For those who already operate cloud compute systems, the main concerns are around wastage and idle costs. In addition there are misgivings about the cost of overall compute power consumption.
No comments:
Post a Comment