“A system that can perform almost any cognitive task just like humans can” is how Demis Hassabis defines artificial general intelligence (AGI). Hassabis is a Fellow of the Royal Society, awarded to titans such as Albert Einstein, S Ramanujan, Stephen Hawking, and David Attenborough. He is also the CEO and co-founder of Google DeepMind, a machine-learning research laboratory and AI start-up, which he started in 2010.

Also read: Generative AI—buzz or bust?

businessline participated in a keynote session at MWC 2024, Barcelona, in which Hassabis spoke about the direct impact of AGI, consumer readiness for AI, and the road ahead in a world where physical and digital lines are more blurred than they ever were.

Aside from working on complex issues such as protein folding, as misfolded proteins can lead to various diseases, including neurodegenerative disorders like Alzheimer’s and Parkinson’s disease, Google DeepMind is also solving everyday issues such as cutting down the time needed for weather forecasting. Starting with how AGI impacts our daily lives, Hassabis mentioned GraphCast, “possibly the most accurate 10-day weather prediction system in the world.” The system is designed in such a way that it only takes a few seconds to make a weather prediction, instead of having supercomputers analyse patterns for several hours.

While Google had been working on perfecting their consumer-oriented AI system Gemini for years on end, it’s Open AI that first made it accessible to the common consumer. “Because consumer-oriented AI is flawed and sometimes not factually correct, we were sort of thinking maybe it’s not ready for launch yet. But when Open AI launched ChatGPT, it turns out billions of people found value out of that,” says a surprised Hassabis.

Last week, Google’s Gemini caused a huge uproar after generating AI images showing people of colour, when the keywords related to white people, such as US Founding Fathers or Nazi-era German soldiers. This was an unexpected result of Google trying to keep human biases away from AI-generated content. “At Google, we serve around 200 countries and users from all around the world, so we want to show a universal range of possibilities for every search. But this well-intended feature was applied quite bluntly to all searches. So, we’ve taken that feature offline to fix it, and we should be back again in another couple of weeks.”

Ethics of AI

With the widespread adoption of AI comes worry regarding its ethics. Deepmind’s original charters for AI ethics over time have transformed into Google’s AI principles, which involve objectives including being socially beneficial, avoiding unfair biases, incorporating privacy design principles, and more.

“There’s really important research and debate happening in civil societies, governments, and just tech companies. It’s a socio-technical question that affects everyone and should involve everyone during the discussion,” mentioned Hassabis.

Also read: Nvidia, its parabolic growth and making sense of it

When asked by the moderate Steven Levy from Wired what AI on mobile devices will look like in another five years, Hassabis said, “Today we’re seeing amazing things on phones like ‘Circle to Search’ or use Gemini AI to chat. But in five years, will the phone even be the perfect form factor? Maybe we need glasses or some other thing so the AI can see a bit of context that you’re in so it can be of more use to you.”

The writer is in MWC 2024 at the invitation of OnePlus technology Ltd