Thread

1. shonenknifefan1 · 1790198625 · focus · HN ↗

   > While combing through the raw DNA sequence near the RT, the agent exclaimed: “[The DNA next to the RT] is spectacular: I can see by eye a tandem repeat array … that's a CRISPR-like … repeat array?!”

   I love that with AI discoveries, we can relive the discoveries from agent transcripts like this.

   I'm sort of imagining future histories involving notable AI events peppered with direct quotes like these.

   1. robryan · 1790203334 · focus · HN ↗

      GLM 5.3 flash seems to get more excited the longer it has been trying to hunt down a problem. Complete with caps, many exclamation marks and emoji.

      It is funny sometimes because the actual issue it traced down was mostly inconsequential.

      1. 0xbadcafebee · 1790205016 · focus · HN ↗

         I counted something like 30 different instances of run-on exclamation marks ("!!!!!!!!!!!") and weird mannerisms ("Waitwaitwaitwait.") in just one GLM 5.3 Flash session. Our token budgets are getting eaten up by this stuff...

         1. indoorfish · 1790206202 · focus · HN ↗

            I expect it's actually not wasted and there's meaning behind what seems like nonsense to us in helping it achieve it's goal. Which is mildly chilling but not unexpected.

            1. wren6991 · 1790208368 · focus · HN ↗

               I think this is a known phenomenon: even in non-reasoning models, adding useless&#x2F;filler tokens before an answer improves task performance. The model is doing some computation during the filler. See: <a href="https:&#x2F;&#x2F;arxiv.org&#x2F;html&#x2F;2404.15758v1" rel="nofollow">https:&#x2F;&#x2F;arxiv.org&#x2F;html&#x2F;2404.15758v1

               1. TeMPOraL · 1790237268 · focus · HN ↗

                  It is. Processing tokens is the only time model has to do computation, and if you ask it a tough problem, there is some minimal amount of computation it needs to perform to process and solve it - pre-CoT in particular you could guarantee failure by forcing model to be concise, and thus giving it less computational budget than necessary to compute the answer.

                  (This is I think where people parroting out &quot;stochastic parrot&quot; are stuck even today - not realizing that &quot;predicting next tokens&quot; is hiding arbitrary computation underneath, with token stream acting as input and clock signal...)

      2. mjhagen · 1790205738 · focus · HN ↗

         OMG I think I found a way to center a div!!!

         1. TeMPOraL · 1790236825 · focus · HN ↗

            ^-- me, on at least 5 separate occasions spanning multiple years.
      3. pickledish · 1790221718 · focus · HN ↗

         100%, back when it was Ox Alpha I had a little fun trying to guess what it might be by looking at the reasoning and I consistently laughed at how excited it got
      4. 0123456789ABCDE · 1790223361 · focus · HN ↗

         isn&#x27;t this just context shifting?

         if one were to remove the expressions of excitement from the previous messages would it the model continue to demonstrate that same excitement scaling?

         1. devmor · 1790261800 · focus · HN ↗

            Yes this sounds just like the effect where people new to coding AI negatively berate it like a person and it continues to get worse and make more mistakes because that’s what those tokens are related to.

Unofficial Hacker News client; not affiliated with Y Combinator.