3 tech pioneers on the biggest AI breakthroughs – and what they expect will come next
As predictions go, they don’t come much bigger.
“This is going to be the most transformational moment, not just in technology, but in culture and politics of all of our lifetimes.”
These are the words of Mustafa Suleyman, Co-Founder of Inflection AI and author of The Coming Wave. He was speaking to the World Economic Forum’s Radio Davos podcast about the impact of artificial intelligence.
Being at the forefront of the development and rollout of generative AI and large language model (LLM) tools, Suleyman has witnessed the transformative power of the technology, and also deeply understands the challenges that lie ahead.
And it’s a sentiment shared by two other leading AI thinkers: Aidan Gomez, Co-Founder and CEO of Cohere and Yann LeCun, Chief AI Scientist at Meta, who, like Suleyman are featured on the TIME100 Most Influential People in AI list.
Here, in an edited version of their interview, the three experts discuss the moments when they were acutely aware of AI’s true potential – and share their thoughts on the future roadmap for the development of the tool.
1. From creating a fictional, Japanese punk band to transforming healthcare
Most people will admit that when they first got their hands on an LLM tool – in most cases, during early 2023 with ChatGPT – their first prompt experiment was something silly. It might have been a request for a comical nursery rhyme or a novel cocktail recipe.
The experience was no different for Aidan Gomez. As co-author of the groundbreaking Attention Is All You Need paper, which he helped to write as a 20-year-old intern at Google Brain, Gomez is credited with changing the AI industry and helping lead to the creation of ChatGPT.
“I got this email from my collaborator who was a co-author, Lukasz Kaiser, on the transformer paper. And it was like, look at this.”
Kaiser went on to show Gomez how he had prompted the technology with a title ‘The Transformer’ and it returned creative fiction chronicling the history of a Japanese punk rock band.
“That was a surreal moment where you saw a machine speaking so fluently that it convinced me I was reading an article about a Japanese punk rock band. And so that was the eureka moment for large language models.”
Looking ahead, Gomez is most excited about the transformative productivity gains generative AI can have on many professions, including healthcare.
“I think doctors spend something like 40% of their working hours writing notes, and if we could give them 40% of their days back to spend with patients, focus on patient outcomes, it means we double the number of doctors effectively overnight.”
AI applications can save European health systems 1.8 billion hours every year, according to a report by Deloitte and commissioned by MedTech Europe, which is the equivalent of having 500,000 additional full-time healthcare professionals available.
How AI can help healthcare.Image: Deloitte
2. The need for an updated Turing Test
The Turing Test – a concept created by British mathematician and computer scientist Alan Turin in 1950 – may need rethinking.
It’s a long-standing test of a machine’s ability to exhibit intelligent behaviour equivalent to a human. But even just the notion that it might need revising was when Mustafa Suleyman truly realized the power of generative AI.
“A modern Turing test would be to try to evaluate whether an AI was capable of acting like an entrepreneur, like a mini project manager and inventor of a new product, to go and market it, manufacture it, sell it, and so on, to make a profit,” he says.
Suleyman believes that generative AI will be able to execute that task within the next decade on an inexpensive, potentially open-source level, which will completely change the economy.
“This is going to be the most transformational moment, not just in technology, but in culture and politics of all of our lifetimes,” he explains. “We’re going to witness the plummeting cost of power.”
3. AI needs to interact with the physical world to become fully realised
“Most of our knowledge in our minds is actually from our interaction with the physical world,” explains Yann LeCun, Chief AI Scientist at Meta and Professor at New York University, who is renowned for speaking out against the hype surrounding AI.
LeCun argues that the short-term outputs created by generative AI tools like text, video, music and image creation have the power to make people more creative and efficient.
However, the next stage will be AI taught not just on the 10 trillion words on the internet that was used to train the current LLMs but on its interaction with the physical world.
“That’s an incredible amount of information because it would take you or me somewhere between 150,000 and 200,000 years to just read it. So it’s unfathomable how big it is,” he says.
“But then you compute how much information gets into the visual cortex by the time a child is four. He or she has seen 15 times more information through vision than the biggest of our LLMs.”
LeCun states that this vast amount of data baked into real life is much bigger than can be found in text – and scientists and engineers do not as of yet know how to emulate this process in machines.
“So we’re still far from having really intelligent systems, but hopefully we’ll have them and then people will interact with those things on a daily basis. They’ll be in their smart glasses, and your interface with the digital world would be an AI system.”
