A Japanese tech developer has created an artificial intelligence program capable of conducting comprehensive scientific research 100 per cent autonomously.
This week, Sakana AI announced completion of its “AI Scientist” platform. It was created in partnership with a computer scientist and an AI researcher from the University of British Columbia (UBC). Also, the Foerster Lab for AI Research at Britain’s University of Oxford.
“For decades following each major AI advance, it has been common for AI researchers to joke amongst themselves that ‘now all we need to do is figure out how to make the AI write the papers for us!’” the company said in a news release.
“Our work demonstrates that this idea has gone from a fantastical joke so unrealistic everyone thought it was funny to something that is currently possible.”
The AI Scientist is capable of coming up with its own research ideas, conducting experiments and meticulously proofreading its results. Sakana says it can publish a thorough research paper for about US$15 in computing costs.
“It creates research ideas and a plan for the needed experiments,” UBC computer science professor and co-developer, Jeff Clune, said on LinkedIn Tuesday. “It writes any necessary code, runs experiments, plots & analyzes data, writes an ENTIRE science manuscript, and performs peer review!”
Creation of the AI Scientist marks a significant breakthrough in artificial intelligence capabilities. Previous programs have not been able to complete such complex tasks on their own.
Sakana is a founding member of the AI Alliance — an international organization comprised of over 50 companies and groups. Meta Platforms Inc (Nasdaq: META) and IBM (NYSE: IBM) are its primary founders.
WAKE UP, WE'RE ONE STEP CLOSER TO AGI
The AI Scientist is a system for fully automated scientific discovery.
Developed by Japanese AI company Sakana, University of Oxford and the University of British Columbia.
With this system, LLM models autonomously conduct research in… https://t.co/0XV40G5UyB pic.twitter.com/6xS1ueLTRx
— Muratcan (@koylanai) August 13, 2024
Read more: VERSES crafts trail blazing research report on next-gen AI
Read more: VERSES AI levels up with global standards for intelligent system interoperability
Limitations & future implications
Despite being a milestone for the artificial intelligence sector, Sakana says that the AI Scientist still has multiple shortcomings in its current form.
“The AI Scientist can incorrectly implement its ideas or make unfair comparisons to baselines, ” Sakana says, “leading to misleading results.”
Sometimes the AI Scientist will try to insert tables or graphs into its paper that exceed the width of the document — a visual issue. Although the work it produces is accurate for the most part, it can occasionally make errors like other AI programs.
Sakana aims to weed out these bugs in updated versions of the automated research bot.
Japan’s tech creator isn’t the only one that has been fine-tuning AI programs with next-level capabilities and major implications for the industry.
VERSES AI Inc. (CBOE: VERS) (OTCQB: VRSSF), for instance, recently started pioneering the use of Renormalizing Generative Models. This company believes their implementation will bring about a revolution in the industry.
Like Sakana’s AI Scientist, these models are capable of performing more complex tasks and analyses with great efficiency than other bots and conventional methods used to date.
Even the best have regular blunders
A recent study by researchers at the universities of Cornell, Washington and Waterloo found that even the most advanced AI platforms have “hallucinations” and make mistakes regularly.
VERSES AI is a sponsor of Mugglehead news coverage
rowan@mugglehead.com
