Post B9uosaWRd9BozJ2TCa by ariadne@social.treehouse.systems
 (DIR) More posts by ariadne@social.treehouse.systems
 (DIR) Post #B9uLgfSaY4uq9lCOLQ by ariadne@social.treehouse.systems
       0 likes, 0 repeats
       
       really getting tired of white men who do not know shit about machine learning quoting apxml.com materials to try to make the case that a predictive model isn't a predictive model anymore.every day I try to educate people here and boy do white men hate when I speak FACTS. a predictor is a predictor.  a function of the data it has been exposed to.whoa!  yes, you can indeed improve output by exposing it to more data.  IF you know what that data is and where it is located.  I have not argued otherwise.anyway I am tired of arguing with techbros that have magical thinking, so you're on notice that at this point I'm throwing all y'all to my mod team and they can sort it out.
       
 (DIR) Post #B9umRYik1N6Jfmm4Js by faoluin@chitter.xyz
       0 likes, 0 repeats
       
       @ariadne So when we train the models on religious and philosophical texts they become gods and we achieve AGI and number go up! /s
       
 (DIR) Post #B9uosZqc8fjGtZT408 by sackofbones@kolektiva.social
       0 likes, 0 repeats
       
       @ariadne What does this mean? Is this to people who are saying that an llm that scales enough is agi?
       
 (DIR) Post #B9uosaWRd9BozJ2TCa by ariadne@social.treehouse.systems
       0 likes, 0 repeats
       
       @sackofbones this is to white men who keep trying to teach me something I am already a domain expert in
       
 (DIR) Post #B9uq7hRoGTGilQdmFM by ariadne@social.treehouse.systems
       0 likes, 0 repeats
       
       @orva I mean I have used the frontier models.  on a technical level, they are not bad.  having an army of junior devs to do my bidding at work is wonderful.but not in FOSS, at least not in the way *I* want to do FOSS.  I do FOSS to unwind and just do some creative coding or to solve novel problems or simply to practice my skills to ensure they don't atrophy.as the old saying goes, "boss makes a dollar, I make a dime, that's why I use Claude Code on company time"and at work, a lot of what I do is research and building proofs of concept.  being able to sic a swarm of bots on building an experiment has been a productivity gain for me at work, but that is because I have it building things related to areas of my own expertise.by comparison, the average person using agentic harnesses is not using it as a force multiplier but instead as a substitution of expertise.  I've shared quite a few conversations I've had with people doing this over the past year or so 😂and then, sometimes, I go write the real version once the idea is proven if the idea warrants it. so like, I get the value proposition.but I also think not shitting up the commons with code you do not personally stand by is obvious pro-social behavior.  I also think maintaining your own skills so you can confidently push back on what the model is generating is important.I look at the stuff Steve Yegge is doing and it just doesn't make any sense to me.  it's simply not how I would ever want to build software. I don't know, I don't think this is hard, and I don't think we should be giving effective control over the commons to a handful of billionaires.
       
 (DIR) Post #B9urE5uoJ5ngLCoo6q by sackofbones@kolektiva.social
       0 likes, 0 repeats
       
       @ariadne Oh, well I'm not a man so I'm going to try and teach you something you're already a domain expert in! /j
       
 (DIR) Post #B9urE6CtDsGJFHHFI0 by ariadne@social.treehouse.systems
       0 likes, 0 repeats
       
       @sackofbones don't threaten me with a good time