iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Mathematicians are not proven to be a vanishing breed of creators. But in an October 1, 2026, Fast Company essay, mathematician Eli Ben-Sasson argues that AI may change who gets celebrated for mathematical discoveries: machines could generate more results, while people increasingly choose problems, steer the work, verify proofs, and explain what the discoveries mean.
What Ben-Sasson means by “the last generation”
The claim is a forecast about status and the shape of mathematical work, not a declaration that human mathematics will end. Ben-Sasson imagines a shift away from the ideal of the lone mathematician who both makes a major discovery and receives the public credit for it. In his view, mathematicians may instead act more like conductors, curators, and interpreters of AI-assisted research.
That distinction matters: guiding a system toward a useful problem and establishing that its answer is valid are still substantial intellectual tasks. The essay’s argument is that the visible source of a candidate discovery could increasingly be an AI, even when human judgment remains essential to the work.
Free tools Windows power users keep installed
One-click scans. No signup required.
Which mathematical examples support the argument?
Ben-Sasson points to examples that he says show AI tackling problems that had resisted researchers for years. These are reports in his essay; the examples and underlying announcements are attributed to him here, not presented as independently verified results.
#1 Best Overall
- Erdős unit-distance problem: The essay says an OpenAI model disproved a conjecture at the center of the problem in May 2026. The question concerns how many pairs of points in a plane can be exactly one unit apart. Ben-Sasson describes the conjecture as having resisted mathematicians for 80 years.
- Ten problems in mathematics and theoretical computer science: The essay says OpenAI announced in August 2026 that its Astra model had resolved or substantially advanced 10 long-standing problems. “Resolved or substantially advanced” is the essay’s characterization; it does not mean all 10 were necessarily solved in full.
- A non-sofic group: Ben-Sasson describes an AI-discovered example of this kind of mathematical structure, which he says mathematicians had sought for 27 years.
- Arithmetic formulas and the permanent: The essay also describes a result establishing a new lower limit on the minimum size of arithmetic formulas for calculating the permanent, a difficult mathematical function.
These cases illustrate why the forecast is plausible to Ben-Sasson: if AI systems can produce useful candidates in difficult areas, the human contribution may move upstream to problem selection and downstream to checking and interpretation. The examples alone, however, do not establish that AI can routinely make discoveries across mathematics or that human creativity is becoming unnecessary.
Why proof verification becomes central
A surprising result is not trustworthy merely because a model states it confidently. Researchers need a way to inspect the reasoning and establish that the conclusion follows. Ben-Sasson says the Astra proofs came with machine-checkable certificates—artifacts that a suitable checking system can validate, even if the proof was not written for a person to follow line by line.
Rank #2
“Every one of Astra’s 10 proofs came with a machine-checkable certificate, because a proof no human wrote is worth something only if we can trust it.”
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchPC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Special offer. See more information about Outbyte and uninstall instructions. Please review EULA and Privacy policy.
That point is about a necessary condition for using machine-produced mathematics, not a claim that certificates settle every question. A checking system can confirm that a formal derivation meets its rules; researchers still need to understand what problem the result addresses, whether the assumptions fit, and what the result means for the wider field. Ben-Sasson expects work on trustworthy verification to become more prominent as machine-generated proofs become more common.
How the roles could change without disappearing
The essay’s implied change is not a clean handoff from humans to machines. It is a redistribution of tasks, and the roles can overlap:
| Part of the work | Human-led discovery | AI-assisted discovery in Ben-Sasson’s forecast |
|---|---|---|
| Generating a candidate result | A mathematician develops and pursues the idea. | An AI may produce a candidate result or proof. |
| Choosing and directing | The mathematician selects questions and research strategies. | People continue to choose problems and steer the system. |
| Checking the result | Researchers examine the argument and its assumptions. | Machine-checkable certificates may help verify formal steps; people assess relevance and meaning. |
| Explaining and receiving credit | The human discoverer is often the visible creator. | Mathematicians may increasingly interpret and communicate discoveries whose candidate solutions came from AI. |
This comparison describes a possible division of labor, not two mutually exclusive kinds of mathematics. A person may work with AI to generate ideas, refine a proof, check it, and explain the result; the proportions may vary by problem and by the reliability of the tools.
Rank #4
What the essay does—and does not—establish
Ben-Sasson is a mathematician as well as the CEO of StarkWare and a blockchain innovator. That background helps situate his perspective, but it does not independently confirm the mathematical examples or prove his predictions about the profession.
Recommended Free Tools
The essay makes a case for taking AI-assisted mathematics seriously, particularly where it reports difficult results and machine-checkable proofs. It does not establish how often such systems will succeed, how broadly the reported methods will generalize, or whether public recognition will actually shift away from human researchers. Those questions remain open. The headline’s “may” is important: the piece argues for a possible transformation, not a settled future.
Quick Recap
Best Value
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
