Apologies for continuing the OT.Sascha Franck wrote: Tue Jan 13, 2026 7:48 am
Dude! f**king Grok has already praised f**king Hitler - and you call it "damn good"?
https://www.theguardian.com/technology/ ... m-x-ntwnfb
For example:
So far I've had damn good/extremely sophisticated results with abstract philosophical questions. E.g. to what extent Wittgenstein's later philosophical views on epistemology and general rejection of metaphysics can be considered compatible with Deleuze's attempt to develop a metaphysics along the lines of Kant's notion of synthetic a priori judgements.
Grok praised Hitler and also gave me a great response to that question.
I'm not sure if you're emphasising one of the two concerns: that the model is trained in a way which reflects and exaggerates Musk's politics, and that LLMs are fundamentally dysfunctional. Both of those can be discussions in the OT forum.
It looks like 2026 is the year in which these things become the main topic of mainstream discussion, and now that I started watching more of this sort of content, YouTube is throwing loads of it at me.
I see that MMT folk such as Richard Murphy and Steve Keen have started talking about it. I notice also that Richard Murphy, as he makes a video about what he sees the likely dire economic repercussions of AI, says that he uses it and finds it useful. Steve Keen mentions that he guesses that it will have huge benefits in the medical realm, but also also frames it in terms of how it can remove labour from the capital/labour dynamic, i.e. similarly to the way Bernie Sanders talks about it, which is more or less that everyone loses except a few rich tech dudes, unless government steps in.
The thing which I'm not yet seeing much discussion about is how much new labour can be created in the designing of tools which use AI. For instance, how many jobs can be created which put AI to specific use for the purposes of eliminating diseases and extending life. There doesn't seem to be much of a limit in terms of the demand for health-related services. What can happen when a lot of resources get reallocated to that?
On top of that, robotics seems to still be at a very clunky state. How much specialised robotics can be created which requires specialised AI software and training, and what percentage of the population can work in this field? Does all that specialisation get subsumed by a few big tech companies, or is there more of a local/small scale movement whereby tools get designed and trained for unique use cases?
How many jobs can be created in the integration of AI with traditional tools, and the constant/ongoing refinement of that integration? What will the state of refinement be in a few years, and what will the software world look like?
You could consider that in relation to audio software. Will the application of AI in the audio realm have the net effect of eliminating jobs, or will the space continue to thrive as AI gets incorporated into tools in ever-more specialised ways?
I have no idea, but 2026 looks like the moment when all these things become the main focus.
And people will have to start making decisions around how useful are AI tools, which to use, whether to run them locally, which companies to boycott, which operating systems to use, who to vote for... all that jazz.