Google DeepMind CEO Demis Hassabis's AI travel had an unexpected start: his early mastery of chess.
Years earlier Hassabis would person the 2024 Nobel Prize successful Chemistry for creating an AI programme that predicted macromolecule structures, he was a kid chess champion who started playing the crippled astatine the property of four. By property 13, he was a chess master competing against adults successful world competitions.
In a lecture earlier this period astatine the University of Cambridge, Hassabis, now 48, explained that chess sewage him "thinking astir reasoning itself" aliases exploring the intelligence processes down analyzable thoughts.
"How does our mind travel up pinch these plans, these ideas?" Hassabis asked. "Perhaps much fascinating to maine than the games I was playing was the existent intelligence process down it."
Related: Google Is Paying $2.7 Billion to Reportedly Rehire an Early Employee Who Built an AI Chatbot Before ChatGPT
Hassabis' first vulnerability to programming arsenic a kid was done an physics chess computer, a beingness committee tin of playing chess against a quality player. Though Hassabis was meant to trial retired different chess strategies connected the computer, he was much willing successful really the machine worked and really personification had programmed it to play chess.
"I retrieve being fascinated by the truth that personification had programmed this lump of inanimate integrative to play chess really good against you," Hassabis said. "I was really fascinated by really that was done and really personification could programme thing for illustration that."
In his early teens, Hassabis began trying to build AI programs himself connected an early location computer, the Amiga 500. From past on, he "was hooked" connected AI and decided to walk his full profession making advances successful the field.
Demis Hassabis. Photographer: Stefan Wermuth/Bloomberg via Getty Images
Hassabis co-founded AI institution DeepMind successful 2010, and it was acquired by Google successful 2014 for more than $500 million. He later invented AlphaZero successful 2017, an AI algorithm that needed only the rules of chess and 4 hours of playing against itself to go the strongest chess subordinate ever, beating quality chess masters.
Hassabis was awarded the Nobel Prize successful Chemistry successful 2024 pinch chap laureate and DeepMind head John Jumper for creating an AI model, AlphaFold2, that accurately predicted the analyzable structures of almost each 200 cardinal proteins, each wrong minutes. The AlphaFold Protein Structure Database, which makes these macromolecule structures freely available, has reached complete 2 cardinal users successful 190 countries, helping pave the measurement for precocious investigation successful areas for illustration Parkinson's treatments and antibiotic resistance.
Related: Google Co-Founder Sergey Brin Is Back astatine the Company 'Pretty Much Every Day.' Here's What He's Working On.
Hassabis pointed retired successful his Cambridge speech that it takes an mean of 10 years, positive billions of dollars, to create a caller drug. According to the London School of Economics and Political Science, the mean costs of processing a supplier ranges from $314 cardinal to $2.8 billion.
In the lecture, Hassabis touched connected the anticipation of processing narcotics much quickly and cheaply utilizing AI, "from perchance years down to minutes and seconds."
Hassabis told DeepMind employees successful London earlier this period that he thinks artificial intelligence will go much intelligent than quality beings wrong the adjacent decade.
English (US) · 
Indonesian (ID) ·