On Mon, Dec 08, 2025 at 08:29:55AM -0500, John Clark wrote:
> On Mon, Dec 8, 2025 at 12:05 AM 'Tomasz Rola' via Everything List <
> [email protected]> wrote:
>
>
> > *>** The Chinese AI DeepSeekMath-V2, is not only the first open source AI
> >> to**win a gold medal at the International Mathematical Olympiad, it also
> >> got a **score of 118 out of 120 points on the Putnam Mathematical
> >> Competition;**3,988 humans took that test and all of them were math
> >> majors at prestigious **universities, but the test was so difficult that
> >> the highest score any **human got on it was 90, and the median score was
> >> zero. *
> >
> >
> > > So, a pretty good problem solver.
>
>
> *Pretty good? A 118 score on the Putnam and just pretty good? The rate of
> improvement is astounding, one year ago the smartest AI in the world
> would've gotten a zero on the **Putnam.*
Yeah, yeah... I wondered what was the whole fuss about this Putnam, so
I browsed a bit and downloaded tasks from last year. I have taken
2024/A6 and 2024/A1, scored max i.e. 24 points for the two (i.e. my
solutions were correct, achieved with pen and paper only and no
looking up in the books etc). Only it took me about ten hours total,
so I would have been disqualified from real P. But I hope I can be cut
some slack, because:
- I am at least twice as old as typical contender
- I took them without preparation, while some (many, all?) prepare
for a year or some (and there is even a textbook for them, for this
specific contest)
- I have to take care of my stuff and so on
- I would not called myself mathematician, or a student of subject,
most of the time, while I felt the affect to maths, maths did not
return the feeling... not too much.
So, anyway, while my performance was not stellar, would I say that
getting score required any kind of superhuman abilities? Nope. It was
quite a fun, however.
> > But I think that a mathematician should be also capable of formulating
> > problems and theories. Consider, for example, Hilbert's problems or
> > Millenium Prize Problems . Would this AI mathematician be capable of
> > inventing Turing machine after being loaded with state of the art as it
> > was in 1936? I doubt it.
>
> *Grigori Perelman got a gold medal at the International Mathematical
> Olympiad in 1982, but he had his 16th birthday only a few days before so he
> was not yet a good enough mathematician to make a significant contribution
> to mathematics, however he kept getting better and a few years later he
> proved one of the Millenium Prize Problems, the Poincare "Conjecture", so
> now it is no longer a conjecture, it is a fact. I think AIs will follow a
> similar, but steeper, trajectory. *
The problem is not about trajectory of some automaton going via
existing road (built by humans, mind you) - this is merely solving a
task of optimisation, so it does not boom into trees. The problem is
about automaton choosing to build a wholly new road. This is what I
expect from a mathematician.
--
Regards,
Tomasz Rola
--
** A C programmer asked whether computer had Buddha's nature. **
** As the answer, master did "rm -rif" on the programmer's home **
** directory. And then the C programmer became enlightened... **
** **
** Tomasz Rola mailto:[email protected] **
--
You received this message because you are subscribed to the Google Groups
"Everything List" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion visit
https://groups.google.com/d/msgid/everything-list/aTgKr5bMkc/7ok71%40tau1.ceti.pl.
