A princess is as old as the prince will be when the princess is twice as old as the prince was when the princess’s age was half the sum of their present age. What is the age of the prince and princess?

The brain-twister can be solved easily by OpenAI o1, a new AI model. Unlike existing large language models, o1 reasons through a problem before delivering the result.

The new model, according to OpenAI, “performs similarly to PhD students on challenging benchmark tasks in physics, chemistry and biology”, besides
appearing promising in areas such as mathematics and coding.

The Sam Altman-led company is offering a preview version of the model through OpenAI’s popular chatbot, ChatGPT, to paid Plus and Team users.

OpenAI said the o1 models outperform existing models such as GPT-4o in a qualifying exam for the International Mathematics Olympiad, where it scored 83 per cent compared with 13 per cent for the latter.

According to The New York Times, the AI model may have the potential to diagnose an illness based on a detailed report about a patient’s symptoms and history. It may also come to the aid of physicists to generate complicated mathematical formulas.

Earlier, OpenAI taught GPT models to imitate patterns from its training data while with o1 it trained the model to solve problems on its own using a technique known as reinforcement learning, which involves a “chain of thought” to process queries.

OpenAI claims the new training methodology makes the model more accurate. “We have noticed that this model hallucinates less,” said OpenAI’s research lead, Jerry Tworek. But “we can’t say we solved hallucinations”.

Such a technology can help computer programmers who use AI systems to write code. In online programming contests known as Codeforces competitions, this new model attained the 89th percentile of participants.

OpenAI is not the only company making steady progress. Google’s Gemini and Anthropic’s Claude are showing nascent reasoning capabilities. In July, Google announced AlphaProof, a project that combines language models with reinforcement learning to solve difficult maths problems.

Yoshua Bengio, a computer scientist at the University of Montreal who has won the prestigious Turing Award, has said that assessing the advances through independent evaluations by scientists and academics, without conflicts of interest, is the way forward.

In a series of posts on X, Noam Brown, a research scientist at OpenAI, said the company was releasing the model in preview now to get an idea of how people use it and where it needs to be improved. The model’s release comes at a time when the San Francisco-based OpenAI is looking to raise billions in funding.

OpenAI has said that “o1 can be used by healthcare researchers to annotate cell sequencing data, by physicists to generate complicated mathematical formulas needed for quantum optics, and by developers in all fields to build and execute multi-step workflows”.

In case you are wondering, the correct answer to the princely problem: The prince is 30 and the princess is 40.