"Massive AI Investment Warned to Trigger Scams Like Cryptocurrency," DeepMind Founder Cautions

"In some ways, it is too exaggerated." Demis Hassabis, CEO and co-founder of Google DeepMind, expressed caution regarding the massive investments pouring into artificial intelligence (AI) following the ChatGPT craze. He pointed out that in this process, the meaning of scientific progress is being overshadowed and that it is leading to overhype and scams similar to what happened with cryptocurrencies.

[Image source=EPA Yonhap News]

원본보기 아이콘

In an interview with major foreign media released on the 31st of last month (local time), Hassabis said that billions of dollars are being poured into generative AI startups and products, adding, "There is overhype and grifting similar to what was seen in cryptocurrencies."

He noted, "Some of the funds that were flowing into cryptocurrencies have now flowed into AI, which is somewhat regrettable," and added, "It obscures the amazing scientific and research achievements." He continued, "In some respects, AI is not exaggerated enough, but in other respects, it is too exaggerated," and said, "We are talking about all kinds of things that are not reality."

Since the launch of ChatGPT, countries around the world have been experiencing an investment boom centered on generative AI. According to market analysis firm CB Insights, venture capital groups invested a staggering $42.5 billion in 2,500 AI startup funding rounds last year. Stock market investors have also flocked to the so-called 'Magnificent 7' tech companies such as Microsoft (MS), Google Alphabet, and Nvidia, continuing the AI-driven stock price surge. In the first quarter of this year, the S&P 500 index rose more than 10%, marking the largest increase in five years.

However, during this process, so-called 'AI wash' cases, where people deceive others by pretending to be related to AI to gain profits, have also been identified. Regulatory authorities have already begun investigations targeting companies making false AI-related claims. Gary Gensler, chairman of the U.S. Securities and Exchange Commission (SEC), warned, "No greenwashing, no AI washing."

Nevertheless, Hassabis emphasized that despite these controversies and concerns, AI is one of the most innovative inventions in human history. He said, "I think we are just scratching the surface of what will be possible over the next decade," and evaluated, "We are at the beginning of a new golden age of scientific discovery, a new Renaissance."

He also cited AlphaFold, the protein structure prediction system developed by DeepMind in 2021, as the best evidence of how AI can accelerate scientific progress. He stated that DeepMind is using AI to accelerate research in biology, mathematics, weather forecasting, nuclear fusion technology, and more, emphasizing that his goal has always been to use AI as "the ultimate tool for science."

DeepMind was founded in London in 2010 with the goal of realizing artificial general intelligence (AGI) that surpasses human intelligence. Hassabis assessed that several major breakthroughs remain before achieving AGI but predicted, "I would not be surprised if it is realized within the next 10 years. About a 50% chance."

He also welcomed ongoing measures such as the follow-up summit meetings between South Korea and France after last year's 'Global AI Safety Summit,' and the establishment of AI safety research institutes in the UK and the US. He said, "This is an important first step," and pointed out, "There is much to be done, and technology is advancing rapidly, so we must hurry."

Regarding the so-called hallucination phenomenon, where generative AI like OpenAI's ChatGPT and Google's Gemini produce incorrect information as if it were factual, he mentioned that various methods are being explored to verify factual evidence through cross-checking and other means. He said, "This is somewhat similar to when AlphaGo was operating," adding, "The network does not just spit out the first move it thinks of. It takes time to think and plans." Recently, DeepMind published a paper explaining a new methodology called SAFE that can reduce errors caused by hallucinations.

© The Asia Business Daily(www.asiae.co.kr). All rights reserved.