Google DeepMind founder returns to campus

Long before artificial intelligence started popping up in general conversations, when it was still considered niche, even bonkers, Shane Legg was captivated by its potential: how we might develop it, and how it might change the ways we live and work.

In fact, the University of Auckland alumni was interested in the topic as a child growing up in Rotorua in the 1980s.

Now he’s one of New Zealand’s most consequential scientists at the forefront of artificial intelligence, and he’s returning to the University where he studied 30 years ago for a public lecture on 4 December, followed by an on-stage discussion with TVNZ journalist Jack Tame.

Shane is co-founder and chief AGI scientist at Google DeepMind, an organisation that’s produced some of the defining breakthroughs of the modern AI era. In his lecture 'AGI Incoming', he’ll explore the questions AGI is now raising for governments, universities, businesses, families, and all of us.

But first, what actually is artificial general intelligence (AGI)?

Shane’s “minimal bar” for AGI is that the AI system must be able to do all the cognitive tasks that people can typically do.

More than 20 years ago, he predicted a 50 percent chance that human-level machine intelligence would be created by 2028, and he hasn’t changed his mind.

“I think we’re getting very close and will probably reach this next year or the year after as some of the remaining cognitive weaknesses of our AI models are getting addressed.

“Meanwhile, they are already clearly superhuman in many areas, from general knowledge and knowing multiple languages, while being extremely capable at things like software engineering and mathematical proofs.”

So, when AGI arrives, what could it mean for everyday life, work and society?

Shane, who co-chairs Google DeepMind’s AGI Safety Council, says purely cognitive work, work that relies on thinking, analysing information and solving problems, will be transformed.

“However, most work isn’t purely cognitive. Furthermore, what generally has a large impact on society isn’t the work of individuals; it’s what teams of people, for example in a company, achieve collectively.

“There are huge opportunities in making various kinds of cognitive work far cheaper and faster, as well as major advances in areas such as medicine and algorithms which could fundamentally improve the quality of human life.”

He also points to the risks, from societal destabilisation due to economic disruption and misuse of AI for cyber and other kinds of attacks, to situations where AI goes rogue, such as the recent hack by an autonomous AI agent into Australia's Medicare portal.

Shane graduated from the University of Auckland in 1997 with a Master of Science in Mathematics. After living in Rotorua and then Hamilton, it was the first time in his life he’d regularly been in the centre of a large city.

“There were so many people, buildings, and events happening each day at the University and in the surrounding city. It was both overwhelming and very exciting. This made me understand that the future for me was going to be in big cities and possibly also living overseas.”

So what advice would Shane give to current or future students considering their options in a time of immense change?

“Times of disruption are also times of opportunity. Keep learning and trying to understand where the new technologies are going, and keep thinking creatively. And when you see a big opportunity, don’t be afraid to have a go at making it happen!”

In his first role after university, Shane worked as a software engineer for a courier company. It was an initial taste of company operations, but he says perhaps the most important thing the role taught him was that it wasn’t what he wanted to be doing.

“I missed the excitement of working on research ideas that were really new and groundbreaking.”

After three years, he quit and joined an AI startup. After that company collapsed during the dot-com crash, and “after the usual Kiwi period of backpacking Europe”, he locked in on AI and embarked on a PhD at the Dalle Molle Institute for Artificial Intelligence Research in Switzerland.

His thesis, which proposed a formal definition of machine intelligence, saw him awarded the Canadian Singularity Institute research prize. He spent a year as a postdoc at the Swiss Finance Institute, building models of human decision-making, followed by two years at the Gatsby Computational Neuroscience Unit at University College London, where he’s now an Honorary Fellow.

It was during his postdoctoral research in London that he met artificial intelligence researcher and entrepreneur Demis Hassabis.

“We decided to start DeepMind in 2010, thinking that powerful general intelligence, or AGI, was going to be one of the most important technological developments in history.”

In 2023, DeepMind merged with Google Brain to form Google DeepMind. As chief AGI scientist, Shane now leads a research group called the DeepMind Institute, where he and his team explore some of the biggest questions surrounding AGI, from machine consciousness and ethics to safety, governance, economics and artificial superintelligence.

They’re questions that would have seemed far-fetched when Shane was a student 30 years ago. Now, as he returns to campus, they're central to conversations about the future of AGI and its impact on society.