Google DeepThoughts claims ‘historic’ AI breakthrough in drawback fixing | Synthetic intelligence (AI)

This web page was created programmatically, to learn the article in its unique location you’ll be able to go to the hyperlink bellow:
https://www.theguardian.com/technology/2025/sep/17/google-deepmind-claims-historic-ai-breakthrough-in-problem-solving
and if you wish to take away this text from our web site please contact us


Google DeepThoughts claims it has made a “historic” synthetic intelligence breakthrough akin to the Deep Blue pc defeating Garry Kasparov at chess in 1997 and an AI beating a human Go champion in 2016.

A model of the corporate’s Gemini 2.5 AI mannequin solved a posh real-world drawback that stumped human pc programmers to change into the primary AI mannequin to win a gold medal at a global programming competitors held earlier this month in Azerbaijan.

In a efficiency that the tech firm referred to as a “profound leap in abstract problem-solving”, it took lower than half an hour to work out methods to weigh up an infinite variety of potentialities in an effort to ship a liquid by a community of ducts to a set of interconnected reservoirs. The purpose was to distribute it as rapidly as doable.

None of the human groups, together with the highest performers from universities in Russia, China and Japan, acquired it proper.

It failed two of the 12 duties it was set, however its total efficiency ranked it in second place out of 139 of the world’s strongest college-level pc programmers. Google mentioned it was a “historic moment, towards AGI [artificial general intelligence]”, which is broadly thought-about human-level intelligence at a variety of duties.

“For me it’s a moment that is equivalent to Deep Blue for Chess and AlphaGo for Go,” mentioned Quoc Le, Google DeepThoughts’s vice-president. “Even bigger, it is reasoning more towards the real world, not just a constrained environment [like Chess and Go] … Because of that I think this advance has the potential to transform many scientific and engineering disciplines.” He cited drug and chip design.

The mannequin is a basic function AI however was specifically skilled to unravel very onerous coding, maths and reasoning issues. It carried out “as well as a top 20 coder in the world”, Google mentioned.

“Solving complex tasks at these competitions requires deep abstract reasoning, creativity, the ability to synthesise novel solutions to problems never seen before and a genuine spark of ingenuity,” the corporate mentioned.

Speaking earlier than the main points had been made public, Stuart Russell, a professor of pc science on the University of California at Berkeley, mentioned the “claims of epochal significance seem overblown”. He mentioned AI programs had been doing properly on programming duties for some time and the Deep Blue chess breakthrough had “essentially no impact on the real world of applied AI”.

However, he mentioned “to get an ICPC question right, the code actually has to work correctly (at least on a finite number of test cases), so this performance may show progress towards making AI-based coding systems sufficiently accurate for producing high-quality code”.

He added: “The pressure on AI companies to keep claiming breakthroughs is enormous”.

Michael Wooldridge, Ashall professor of the foundations of synthetic intelligence on the University of Oxford, mentioned it gave the impression of a powerful achievement and “being able to solve problems at this level is exciting”. But he questioned how a lot computing energy was wanted. Google declined to say, other than confirming it was greater than obtainable to a mean subscriber to its $250-a-month Google AI Ultra service utilizing the light-weight model of Gemini 2.5 Deep Think within the Gemini App.

Dr Bill Poucher, govt director of the ICPC, mentioned: “Gemini successfully joining this arena, and achieving gold-level results, marks a key moment in defining the AI tools and academic standards needed for the next generation.”

skip past newsletter promotion

Four machine intelligence breakthroughs

1957 The Perceptron

Frank Rosenblatt, an educational at Cornell University,labored out that it ought to be doable to create a “perceiving and recognising automaton”. He named it the Perceptron and said an digital system would be capable of study to recognised patterns in optical, electrical or tonal data “in a manner which may be closely analogous to the perceptual process of a biological brain”. The following 12 months he constructed the gadget, which was the scale of a small room. It was thought-about one of many early breakthroughs in synthetic intelligence based mostly on neural networks.

1997 Big Blue

In May 1997, IBM’s Big Blue turned the primary pc system to defeat a reigning world chess champion in a match underneath commonplace match controls. It beat Garry Kasparov in what turned an inflection level in computing energy, however the contest was shut. Kasparov gained the primary recreation, Deep Blue the second adopted by three attracts. Deep Blue gained recreation 6 to safe the win. It confirmed how brute power computing energy may create a system to defeat a human, albeit at a slender process. “The computer is far stronger than anybody expected,” mentioned Kasparov, conceding defeat.

2016 AlphaGo

Go is without doubt one of the most advanced video games ever devised, and one of many world’s grasp gamers was Lee Sedol, a South Korean skilled. In 2016, DeepThoughts, the UK AI firm arrange by Demis Hassabis, took him on with its pc AlphaGo. It gained 4-1 and a few of its strikes appeared to show really unique considering. Move 37 specifically went down in lore. Hassibis said: “It might be the first glimpse of a bright and bold future where humanity harnesses AI as a powerful new tool, helping us discover new knowledge that can solve some of our most pressing scientific problems.”

2020 AlphaFold

Another breakthrough by Hassibis and DeepThoughts was an AI program that may predict how proteins fold into 3D shapes, a extremely advanced course of elementary to understanding life’s organic equipment. The Royal Society, the 360-year previous London scientific establishment, referred to as it “a stunning advance”.

When researchers understand how a protein folds up, they will begin to uncover mysteries similar to how insulin controls sugar ranges within the blood or how antibodies struggle viruses. After additional iterations, the system helped Hassibis and his colleague John Jumper share a Nobel prize for chemistry in 2024.


This web page was created programmatically, to learn the article in its unique location you’ll be able to go to the hyperlink bellow:
https://www.theguardian.com/technology/2025/sep/17/google-deepmind-claims-historic-ai-breakthrough-in-problem-solving
and if you wish to take away this text from our web site please contact us

Leave a Reply

Your email address will not be published. Required fields are marked *