Artificial intelligence is advancing at unprecedented speeds. This weekend, OpenAI – the company behind ChatGPT – presented Astra, a new AI model capable of solving problems that, until now, were beyond human intelligence. According to its CEO, Sam Altman, the new tool can solve ten mathematical problems that had remained open for years. Among them are two of the so-called Erdős problems, posed by the Hungarian mathematician Paul Erdős, one of the most recognized figures in the discipline, who left dozens of open problems since his death in 1996.
Read more New complaint against Judge Peinado, this time filed by Manos Limpias
The ten problems belong to areas of high complexity: high-dimensional geometry, coding theory, operator algebras, and quantum complexity, among others. OpenAI took advantage of the presentation to announce a new program aimed at promoting scientific research for 100,000 scientists and mathematicians, offering them its models for free. “We want to empower scientists and mathematicians with tools that accelerate discovery,” the OpenAI statement argued.
Decades of mathematical problems, solved with AI for just 2,000 dollars
OpenAI has even calculated the cost of solving these operations with its new AI model. According to the company, 2,000 dollars are enough to solve these mathematical puzzles if the token rate, the internal payment currency of the AIs, is used.
On the other hand, in early June, the International Mathematical Union (IMU) published a statement warning about the risk of incorporating AI tools without sufficient regulations or guarantees in solving mathematical problems. The document, known as the “Leiden Declaration on Artificial Intelligence and Mathematics,” has already gathered more than 3,400 signatures from academics, mathematicians, and professors from some of the best international universities.
Also read
The International Mathematical Union calls for more regulation and support for public research
Furthermore, the text raised a “serious ethical concern at a time when the consequences of large-scale investment in artificial intelligence are being widely debated regarding war, mass surveillance, political disruption, and environmental damage.” “If we do not act, we risk being complicit in supporting technologies that threaten much more than the practice of mathematics,” it concluded.
Read more Massimo Dutti sneaks into the top 10 most desired brands in the world
To address these risks, the statement outlined a series of recommendations for individual mathematicians – including affirming human authorship of work, assessing its ethical consequences, and using available tools responsibly – as well as advice for mathematical organizations and nonprofit research funders – urging them to maintain standards of rigor, protect copyrights, and strengthen support for public laboratories.
Likewise, the document called on policymakers for regulation of the artificial intelligence industry and more investment in “public computing infrastructure.”
The new model incorporates “multi-agents” that collaborate to solve complex problems in a coordinated way
On the other hand, the new ChatGPT model incorporates innovations in the way artificial intelligence operates. Astra will be a model adapted to more complex tasks. The AI tool will feature a multi-agent system, that is, instead of functioning as a single assistant that answers a question step by step, it will be able to coordinate different specialized processes working in parallel.
In an academic research context, this can allow dividing scientific problems into subproblems and autonomously reviewing their own results. A functioning that approaches the work of a team of researchers, since each AI agent fulfills a specific role or function to execute a task.
Read more Lost in a subway station
