Some mathematicians are claiming a major breakthrough thanks to ChatGPT. A recent case has brought attention to how AI can assist in solving complex math problems that have stumped experts for decades. While AI solutions in mathematics are often unreliable, this new development has sparked excitement and caution across the field.

A Surprising Solution to a Long-Standing Math Puzzle

Earlier this month, a 23-year-old named Liam Price shared a solution to one of the Erdős problems, a set of notoriously difficult mathematical conjectures left by mathematician Paul Erdős. Remarkably, Price, who doesn’t have an advanced math degree, used ChatGPT to generate his answer by prompting the AI with a specific question. Many solutions from AI in similar cases have been inaccurate or incomplete, but experts who reviewed Price’s response believed it was genuine and potentially groundbreaking.

Leading mathematicians, including Terence Tao from UCLA, analyzed the solution and noted that the AI approached the problem differently than previous attempts. Instead of following traditional methods, the AI applied a well-known formula in a novel way, which human mathematicians hadn’t considered. This unexpected approach could hint at new ways to think about large numbers and their properties, sparking hope for future discoveries.

The Role of Human Experts and Caution Ahead

Despite the excitement, experts emphasize that AI still needs human oversight. The raw output from ChatGPT often requires careful review and interpretation. Jared Lichtman, a mathematician at Stanford, explained that an expert’s input was essential to understand and validate the AI’s proof. In many cases, AI solutions have simply rediscovered old results or provided proofs that contain errors.

While this particular breakthrough is promising, some caution remains. Past claims of AI solving difficult problems have turned out to be false or overstated. For example, an OpenAI vice president briefly announced that ChatGPT had solved another Erdős problem, only to retract the claim after scrutiny. This highlights the importance of careful verification before declaring major breakthroughs based on AI-generated results.

Overall, the recent success demonstrates that AI can sometimes think outside the box, but it still relies heavily on human experts to refine and confirm findings. The ongoing collaboration between AI tools and mathematicians could open new doors, but it’s too early to tell how significant these discoveries will be in the long run.

Inspired by

• https://futurism.com/artificial-intelligence/mathematicians-claim-significant-discovery-using-chatgpt

Sources

• scientificamerican.com
• erdosproblems.com