People are inclined to pay attention when the co-founder of Hugging Face, a $4.5 billion artificial intelligence start-up, shares his thoughts on the direction of science. Thomas Wolf, however, does not agree with the evangelists. Rather, he has raised questions about whether the AI systems being developed today would bring about the significant scientific advancements that some of their proponents prefer to think about. The Hugging Face guru explains why chatbots are not perfect The comments made him to go against other AI industry prominent figures, with Dario Amodei of Anthropic and Sam Altman of OpenAI say notable and significant change is closer than before. In a blog, Amodei suggested that AI has the possibility to compress centuries of biological data into a 10-year study. Wolf is not convinced that AI will be significant and according to him, scientific breakthroughs must be the kind of innovations that make our understanding of the world much easier. He argues that AI models of today are not at par with Copernicus who suggested that the sun and not the earth was the centre of the solar system. Wolf clarified the two issues, starting with ChatGPT and similar systems that tend to align with the person using them. “When you ask a question, the AI you are using hypes you through describing it as interesting or important, and this kind of hyping as much as it is good, does not usually lead to scientific thought revolution.” Wolf. Secondly, the next word in a sequence is what these massive language models are being trained to guess and anticipate and they intentionally produce predicted results rather than the unexpected ones. On the other hand, serious paradigm shift scientists seldom choose the easiest route. “The scientist is not trying to predict the most likely next word. He’s trying to predict this very novel thing that’s actually surprisingly unlikely, but actually is true,” Wolf said. The Hugging Face co-founder has been thinking about this topic for the last few months. His interest was sparked after he read an essay penned by Anthropic’s Amodei, who posited that “AI-enabled biology and medicine will allow us to compress the progress that human biologists would have achieved over the next 50-100 years into 5-10 years.” Startups are pursuing their goals Wolf acknowledges that he has been thinking about the topic for months and he grew less convinced the more he flipped Amodei’s forecasts around in his mind. He sees a more humble reality, where AI might be a supporting tool rather than a game-changer. AlphaFold, an AI system developed by Google’s DeepMind that mapped protein structures and created new avenues for drug discovery, caused a stir. According to Wolf, AI is just a support acts even though it is super advanced, and they facilitate speedier material navigation, but they cannot replace the bizarre creative leaps that define Nobel-caliber discoveries. All this talk by Wolf has not discouraged new businesses from entering the stage, and some organizations such as FutureHouse and Lila Sciences have promised to apply AI in their innovation processes. However, it remains an observatory period of how many will fulfill those promises. The negative thoughts and opinions by Wolf are a good counterbalance to those who only shoot positive predictions in the industry. According to him, AI has the capacity to revolutionize laboratory work and spur small steps forward, but when it comes to rewriting the rules of knowledge itself, he believes the burden still rests on human shoulders. A recent article by Cryptopolitan also shows how Silicon Valley also strongly believes that AI can help humans with the gift of near immortality, making humans live up to 200 years. Sharpen your strategy with mentorship + daily ideas - 30 days free access to our trading program
