Karl Fredrik đŚ | đ WHY âď¸ QFOX
wohaliEven the AI lovers are getting disillusioned. good
https://www.thealgorithmicbridge.com/p/im-losing-all-trust-in-the-ai-industry
Neville Park@wohali for those who don't want to give Substack the traffic https://archive.is/ITkpj
matthew green@wohali "A disturbing amount of effort goes into making AI tools engaging rather than useful or productive.
I don't think this is an intentional design decision."
oh, honey
I don't think this is an intentional design decision."
oh, honey
Steven Heywood@wohali
Thanks for this. Essentially it boils down to: "There are some tools that are potentially useful but their development is neglected because the companies and developers are making big money promising investors they can have unicorns tomorrow and putting about wild stories to make that promise plausible. In the meantime we are left with Microsoft Clippy's bigger, more actively unreliable cousin and tools for creating interactive pictures of girls with unfeasibly large breasts and seven fingers and it's the latter that will get the work done on it because it generates more clicks and income."
Thanks for this. Essentially it boils down to: "There are some tools that are potentially useful but their development is neglected because the companies and developers are making big money promising investors they can have unicorns tomorrow and putting about wild stories to make that promise plausible. In the meantime we are left with Microsoft Clippy's bigger, more actively unreliable cousin and tools for creating interactive pictures of girls with unfeasibly large breasts and seven fingers and it's the latter that will get the work done on it because it generates more clicks and income."
ouilaguerre@wohali most infuriating clown skit
DeltaWye@wohali âHow did it happen? OpenAI canât tell you. How can it be prevented? OpenAI canât tell you. Because OpenAI doesnât know. No one does. AI models behave weirdly, and as weird as their behavior is, their misbehavior is weirder. When you manage to jailbreak a model or someone else prompt-injects it, what happens next is unpredictable. If anyone can lure ChatGPT into roleplaying something it shouldnât, then it is inherently not a safe product.â
This is an incredibly fatal flaw which means you totally cannot use LLMs for most, if not all, professional applications without exposing yourself to incredible risks that even the dullest humanâs common sense would recognize.