Goran Rakocevic

[Goran Rakocevic @rgoran8]

@kevinnbass So, just directly generate binaries or something? Not sure we are there yet.

[Goran Rakocevic @rgoran8]

@naval If no one had nuclear weapons we would be fighting and dying en masse in WW7 by now.

[Goran Rakocevic @rgoran8]

@DavidRBellamy Also, smallpox virus genome is public. So are ebola and nipah.

[Goran Rakocevic @rgoran8]

@tdietterich I don't have a great solution, but all of these feel like they are on the wrong track, which imo complicates life for anyone wishing to do genuine science.

aiXiv seems least problematic, but I don't sure I follow why no human author?

[Goran Rakocevic @rgoran8]

@sriramk I would say that prize problems are not analogous to creating a cancer cure, but a somewhat useless yet very specific chemical, such that having a deep understanding of the process of its discovery gives us tools to build cancer cures in future.

[Goran Rakocevic @rgoran8]

@sytelus > They took it with grace, excitement and without whining about why their role is still important.

Have we been on the same https://t.co/mJ01Zq5a9l?

[Goran Rakocevic @rgoran8]

@ronrule Why would you want to though, just use the service?

[Goran Rakocevic @rgoran8]

@ChrisGPT There is something off, or maybe a specific use case? I have been running 2 to 3 threads, one often on ultra and I rarely see below 30% left. They usually do not run through the night though, so if it's 2-3 days of 24h, that might explain it.

[Goran Rakocevic @rgoran8]

@MaleManlpulator Lead, not sole.
But yes, this is very strange to me to. Maybe there is some field-specific precedent for this?

[Goran Rakocevic @rgoran8]

@sama One thing I do not get: How and why would you consider it appropriate for Tristan to be the lead author on something he did not actually do? The work concurrent and independent as I understand it.
Clarifying that up would be a huge step to clearing up this mess.

[Goran Rakocevic @rgoran8]

@OpenAI Y'all are sure this new model did not figure out it's easier to hack into your own servers and look at Buckmaster's chats?

[Goran Rakocevic @rgoran8]

@RhysSullivan I actually like this, Sol just went ahead and did stuff way more often than I wanted it to, so I had to keep explicitly telling it to just respond.

[Goran Rakocevic @rgoran8]

@ibgcvc It kind of always was maintain vs. buy, but people usually only found out after building...

[Goran Rakocevic @rgoran8]

@dggoldst Nah, things will settle on author is responsible, does not matter how it was written.
It will take a generation of AI much less likely to produce gibberish, and a handful of seminal papers co-authored with AI.
The whole debate will feel silly overblown (which it isn't atm)

[Goran Rakocevic @rgoran8]

@sayan_lakhoua @Apple Up until you need to plug it into a crudely manufactured/high pin tension outlet somewhere. I have bent the pins on the classic one doing that.
How long will this mechanism last?

[Goran Rakocevic @rgoran8]

@curiouswavefn Is the problem that you are not fully "equipped" to understand it (i.e. someone stronger in maths should be able to follow), or is the proof highly convoluted and cryptic, even for a maths person.
If former, please find a coauthor.

[Goran Rakocevic @rgoran8]

@Quasilocal What exactly happens when the review is in? It gets passed to the AI?! That paper needs a coauthor, not a reviewer.

[Goran Rakocevic @rgoran8]

@karololszacki @LLMJunky @stolinski Not for 0$, with 0$ penalty. So, you pay 12299 - 9288.

[Goran Rakocevic @rgoran8]

@Andrew_Akbashev I do not see this as sign of people outsourcing their thinking. It's worse: very little think left once it's writing time (or, at least that being the major perception).

[Goran Rakocevic @rgoran8]

@TokenGremlin 4) We need to be seen as making sacrifices that cost the business, in case someone does manage to abuse our model, we lose less face.