![](\"https:\/\/i2-prod.dailystar.co.uk\/incoming\/article31686096.ece\/ALTERNATES\/s615b\/0_Pensive-thoughtful-contemplating-caucasian-young-man-thinking-about-future-planning-new-startup-loo.jpg\")
<\/p>\nAI is now cleverer than humans, according to boffins.<\/p>\n
Researchers have made the world's first scientific discovery using a language model which is the tech behind chatbots like ChatGPT. Google\u2019s DeepMind was able to solve puzzles using solutions that go beyond the best that human mathematicians have ever come up with.<\/p>\n
Pushmeet Kohli, head of AI for science at DeepMind, said: "When we started the project there was no indication that it would produce something that's genuinely new. This is the first time a new scientific discovery has been made by a large language model.''<\/p>\n
READ MORE: Former butler lifts lid on William and Kate's secret trips to boozer and Tesco<\/b><\/p>\n
For the latest weird and wild news stories from the Daily Star, click here.<\/i><\/b><\/p>\n
Boffins are probing if large language models – which underpin modern chatbots – can do more than repackage information learned in training. The models – aka LLMs – are powerful neural networks that learn the patterns of language, including computer code, from vast amounts of text and other data.<\/p>\n
<\/p>\n
Since the arrival of ChatGPT last year the technology has debugged faulty software and churned out everything from college essays and travel itineraries to poems about climate change in the style of Shakespeare. But while the chatbots have proved popular they do not generate new knowledge.<\/p>\n
To build "FunSearch", short for "searching in the function space", DeepMind harnessed an LLM to write solutions to problems in the form of computer programs. The LLM is paired with an `evaluator' that ranks the programs by how well they perform.<\/p>\n
The best programs are then combined and fed back to the LLM to improve on. This drives the system to evolve poor programs into more powerful ones that can discover new knowledge.<\/p>\n
![](\"https:\/\/i2-prod.dailystar.co.uk\/incoming\/article31686102.ece\/ALTERNATES\/s615b\/0_GERMANY-US-INTERNET-AI-ARTIFICIAL-INTELLIGENCE.jpg\")
<\/p>\n
The researchers set FunSearch loose on two puzzles. The first was the cap set problem.<\/p>\n
It deals with finding the largest set of points in space where no three points form a straight line. FunSearch churned out programs that generate new large cap sets that go beyond the best that mathematicians have come up with.<\/p>\n
The second puzzle was the bin packing problem which looks for the best ways to pack items of different sizes into containers.<\/p>\n
While it applies to physical objects, such as the most efficient way to arrange boxes in a shipping container, the same maths applies in other areas such as scheduling computing jobs in data centres.<\/p>\n
For the latest breaking news and stories from across the globe from the Daily Star, sign up for our newsletter by clicking <\/i><\/b>here<\/i><\/b>.<\/i><\/b><\/p>\n