On 14/10/2023 14:11, Quadibloc wrote:
> On Saturday, October 14, 2023 at 12:58:03 AM UTC-6, RichA wrote:
>> https://phys.org/news/2023-10-supernova-ai.html
>
> If astronomy stopped when supernovae were found, this would be a
> problem. Since, instead, detecting supernovae is just the first step,
> and the reason we want to find them is so that we can observe them
> in detail to be able to draw scientific conclusions about them...

Ironically supernova watch is still something where amateurs contribute
to professional surveys since there are so many galaxies to monitor.
Once an SNR is reported bigger professional scopes are turned on it -
catching the light curve on the rise is very important.

It is far more advanced for asteroid detection for planetary defence and
comet hunting since that has already been automated. The days are long
gone when most comets have human discoverers names on them. A human
discovered comet is now something of a rarity.

Professional astronomers are mostly astrophysicists and spend only a
tiny proportion of their time actually observing. The rest of the time
is spent preparing the observational programme and interpreting the data
that results from it. Observing itself is less than 5% of all the time
spent. You can only get big telescope time for sensible proposals...

Technical staff tend to operate the telescopes for astronomers in the
bigger facilities since there isn't time to waste learning to drive one!

AI will probably help weed out weak observing programme though.

> the threat of machines replacing astronomers is still quite distant.
>
> I mean, if this AI stuff were really any good, it would have proved the
> Riemann hypothesis by now!

Remember they claimed the four colour map theorem quite a long time ago.

It is not beyond the realms of possibility that Google's zero AI will be
able to break some previously intractable mathematical conundrum before
too much longer. A combination of brute force and directed attack with a
tiny amount of human guidance is the most likely route to success.

The AI Go programs have found novel defensive structures that have never
previously occurred in all of recorded human play and to everyone's
surprise have become stronger than the very best humans at Go.

--
Martin Brown