The AI revolution is 'unprecedented' in the scale of human history, new report finds
A new report from Mary Meeker, dubbed the "Queen of the Internet," shows that the adoption of artificial intelligence (AI) is the fastest technological change in history.
The growth of artificial intelligence (AI) in our lives is officially unprecedented, according to the woman dubbed the "Queen of the Internet".
Mary Meeker, a venture capitalist known for her internet trend insights, released a 340-page report titled 'Trends - Artificial Intelligence,' where copious research and charts show what she characterises as the “unprecedented” pace of change in which AI is being invested in, developed, adopted, and used.
Meeker ran capital firm Kleiner Perkins’ growth practice during the 2010s, where she invested in future tech giants like Facebook, Spotify, and Canva.
A profile in Forbes credits her with predicting the rise of Apple, Google, and the digital economy with her previous trend reports.
“To say the world is changing at unprecedented rates is an understatement,” Meeker, now the co-founder of venture capital fund Bonds, writes in the new report’s introduction.
ChatGPT is history's 'biggest overnight success'
One of the staggering charts of the report shows that ChatGPT, OpenAI’s AI chatbot, reached 800 million users by April 2025, a couple of years after its initial launch in October 2022.
The company’s revenue also skyrocketed in a similar fashion; from zero in 2022 to just under $4 billion (€3.5 billion) by 2025.
Based on user data, the rise of ChatGPT, in particular, is history’s biggest “overnight success,” nine years after the company was founded in late 2015, Meeker wrote.
"And, unlike the Internet 1.0 revolution, where technology started in the USA and steadily diffused globally - ChatGPT hit the world stage all at once growing in most global regions simultaneously," according to the report.
The report shows that ChatGPT reached 100 million global users in less than 2 months after its launch, the fastest technology to do so.
In comparison, it took Facebook 4.5 years to reach the same number of users after its launch in the early 2000s.
Meeker estimates in the report that it will take three years for a majority of households to adopt AI technology, down from the 12 years it took for households to start using desktop internet regularly.
Companies being 'extremely aggressive' with AI development
There are various factors at play for the “rapid and transformative” rise of AI, Meeker writes in the report.
There’s general buy-in from new AI company founders and more traditional companies for AI adoption, Meeker contends, demonstrated by cash flows being "increasingly directed" towards AI "in efforts to drive growth and fend off investors".
The report notes that technology’s biggest players, including NVIDIA, Google, Meta, Microsoft, and China’s Baidu have all increased the mentions of AI in their corporate earnings reports to shareholders since 2022.
That, paired with a new wave of AI company founders that are "extremely aggressive" with all stages of the AI product development - from innovation, to product releases, acquisitions, cash burn, and capital raises - means that AI “user and usage trending is ramping materially faster” than before.
The report also noted that the number of new AI models has gone up 167 percent year over year since 2020, and the size of the data sets they are using is up 260 percent in the same period.
Meeker notes that the decrease in costs to develop new models is also unprecedented.
Citing Stanford research, the report shows inference costs for those that use the tech have dropped 99 per cent over two years, even though the cost of training a model is up to $1 billion dollars (€850 million).
The pace at which competitors can match each other on the market is unparalleled. For instance, Meeker notes that NVIDIA’s 2024 Blackwell GPU chip, which helps train AIs to do what users expect, has 105,000 times less energy per token than the 2014 Kepler model.
"It’s a staggering leap, not just of cost reduction, but of architectural and materials innovation that is reshaping what’s possible at the hardware level," she writes.
When AI gets it wrong: Overconfidence
mirrors human brain condition
By Dr. Tim Sandle
June 10, 2025
Image: © AFP Josep LAGO
AI is still relatively immature and there remains plenty to iron out in the developmental phrase. There is also much to be done with the way humans interact with AI systems. One current predicament lies with how much faith should we be putting into AI?
Agents, chatbots and other tools based on AI are increasingly used in everyday life by many. So-called large language model (LLM)-based agents, such as ChatGPT and Llama, have become impressively fluent in the responses they form. However, many provide convincing yet incorrect information.
University of Tokyo researchers have drawn parallels between this issue and a human language disorder known as aphasia, where sufferers may speak fluently but make meaningless or hard-to-understand statements.
Aphasia is a communication disorder that results from brain damage, typically to the language centres in the left side of the brain. This damage can make it difficult to speak, understand, read, and write.
According to lead researcher Professor Takamitsu Watanabe from the International Research Center for Neurointelligence (WPI-IRCN): “You can’t fail to notice how some AI systems can appear articulate while still producing often significant errors… But what struck my team and I was a similarity between this behaviour and that of people with Wernicke’s aphasia, where such people speak fluently but don’t always make much sense. That prompted us to wonder if the internal mechanisms of these AI systems could be similar to those of the human brain affected by aphasia, and if so, what the implications might be.”
The researchers used a method called energy landscape analysis, a technique originally developed by physicists seeking to visualize energy states in magnetic metal. This approach has recently been adapted for neuroscience. The scientists examined patterns in resting brain activity from people with different types of aphasia and compared them to internal data from several publicly available LLMs.
This led to the discovery of some striking similarities. The way digital information or signals are moved around and manipulated within these AI models closely matched the way some brain signals behaved in the brains of people with certain types of aphasia, including Wernicke’s aphasia.
It is hoped the research will lead toward better forms of diagnosis for aphasia or provide insight to AI engineers seeking to improve LLM-based agents.
AI for example can sometimes become locked into a kind of rigid internal pattern that limits how flexibly they can draw on stored knowledge. Understanding these internal parallels may be the first step toward smarter, more trustworthy AI.
The research appears in the journal Advanced Science, titled “Comparison of Large Language Model with Aphasia.”
No comments:
Post a Comment