"Can A.I. ameliorate the inequities of our world other than by pushing us to the brink of societal collapse?
"If A.I. is as powerful a tool as its proponents claim, they should be able to find other uses for it besides intensifying the ruthlessness of capital.”
Google “We Have No Moat, and Neither Does OpenAI” - Google insider admits that Open Source #AI is already far ahead of anything Google, Microsoft or any other private corp is working on. Not surprising. Release exciting, high-potential new technology into the wild and this is what happens. So Google asks: Where do we add value to the process? Good question. Or is this a case where legacy industry giants simply won’t have a role? #OpenSource is leading the way, and that’s both encouraging AND somewhat alarming. Alarming because the potential for negative outcomes, fraud, misinformation and massive job loss remain in the absence of government oversight and regulation. #ArtificialIntelligence#LLM#ChatGPT#GPT4#singularityhttps://www.semianalysis.com/p/google-we-have-no-moat-and-neither
@CharlieMcHenry Well Altman said AI would become a commodity in short order. He said it’s all about distribution. Partnering with Microsoft was massive for them. For both of them. I would expect MS will reign in the wilder excesses of the new technology. MS take reputations risk seriously and have a lot to lose. My two cents.
@mnl as far as what that means for "real challenges", here's a short list of things that are nearly impossible to automate, everyone runs into them, and nobody has good ways to avoid it because it's too complex to "start off" with these patterns:
going from 1 to many (resource, env, region, AZ, etc)
@mnl none of the existing stuff out there really solves any of that imo. What the config lang stuff seems to have (value wise) is making it easier to write reams of boilerplate from a simple config that can be type checked. And yeah, while it's valuable to be able to have a deploy.yml file with
I'm intrigued by @simon's exploration of #LLM prompt injections.¹ It reminds me #GEB Contracrostipunctus — If Record Player X is sufficiently high-fidelity, then when it attempts to play the song 'I Cannot Be Played on Record Player X', it will break.² And jumping out of the system with "Record Player Omega" that works by scanning the record and reconfiguring the player into something which won’t break.³
I wonder if future #LLM-driven IDEs will look more like GitHub than a code editor: you file an "issue" to tell the bot what you need, it creates a PR for you to review, you test it, review it, make some changes, ask it to change other things and refine your specs in the issue until you're happy and merge.
That's the workflow I find myself using with ChatGPT, but all the copying and pasting between browser and IDE, and repetitions in the chat make it clumsy. GitHub solved most of the UX for that already.
Little more hands-on with a certain #LLM for some time now.
Once I learned what the "stop sequence" is actually good for, my instinctive ascription of at least a little bit of personality to the thing disappeared immediately.