Die Gegenreaktion der KI hätte zu keinem besseren Zeitpunkt kommen können
https://www.infoworld.com/article/3626533/the-ai-backlash-couldnt-have-come-at-a-better-time.html
Die Gegenreaktion der KI hätte zu keinem besseren Zeitpunkt kommen können
https://www.infoworld.com/article/3626533/the-ai-backlash-couldnt-have-come-at-a-better-time.html
5 Kommentare
The article hits some points that reflects my issue with the current put AI into everything and use it as the main selling point. AI like an operating system needs to be boring and not distract or interrupt me when I’m focused on something. I need to be able to decide when and how and if to use these things.
the tech companies have oversold a technology that is not ready for prime time ; bubble ready to burst
I like Scott. He is pragmatic, and that, more than hype, is what the field require. It might not want it, but it is what it needs. This article touches on a few things he and I discussed a few months ago. The heavy-handed „you *shall* use it“ will typically generate a backlash. When it’s actually working and tailored for specific purposes – that’s when it becomes useful. There’s studies that show this. Generic and broad (and error-prone) is a hindrance, not a help.
We need the bubble to burst to get there, to get rid of the snake-oil salesmen that make grandiose promises they’ll never deliver on and who burn through cash like there’s no tomorrow.
RamaLama is a good idea, just like containerising an application for easy delivery, you can tune and tweak an LLM for a specific purpose and containerise it. I use ollama, but that’s because I’ve pulled the models I’ve tinkered with *for me*. That doesn’t scale if you need to do it for 500, 5,000 or 500,000 employees.
Scott puts it well when he say: „it needs to be boring“. When it is just another tool you can reach for as and when you need to.
People should at least boycott ChatGPT to send a message to Sam Altman for fucking up the DRAM & NAND Pricing Markets and quickly hasten the OpenAI Circle Jerk Bubble.
I used to be real excited and defended a.i. alot, but it’s hard to defend it now.
Ram prices are crazy.
Constantly being asked to upgrade my google speakers to Gemini, even though it’s less capable right now and I’ll have to pay a subscription to talk to it longer.
We have a subscription to chatgpt at work, but it’s been giving me fake answers lately or stringing me along to eventually not being able to do what it said it could. It’s the paid version, it’s already got it’s money, no need to be like this.
Newer versions of software we use at work advertise „new a.i.“ assistant. Excited to maybe get it to help me with previous version bugs, only to realize this isn’t a.i., it’s just a fancy help menu.
Yeah I’m done with it all now.