There will always be a need for both human oversight and accountability, and this is a good example. I think the net result will be, eventually, more and better jobs. It's a better job to validate the transcriptions than to actually transcribe.
Another example in medicine, radiologists will start handling orders of magnitude more cases. But the number of scans done might also increase exponentially as costs likewise drop.
I feel there's some thread that links the USPTO, Treasury, and FTC on guaranteeing the integrity of the dollar and commerce. We don't have a gold standard and the OPEC cartel is a weak substitute. It is some basis for the dollar that in the year 2020 an Intant Pot costs $150 and is designed and manufactured to certain specs and quality. If you look across the economy at all examples of this perhaps it's possible to create a new peg for the USD. Maybe that could even be accomplished without "blockhain" or "AI" :) and just USPS mail-in satisfaction surveys/warranties of some sort tied to corporate lending/insurance.
Look for a discounted Lenovo ThinkPad that you can test out in store to make sure it has a sturdy keyboard and can handle key presses without a hollow feeling.
I'd recommend trying to install Alpine Linux as a challenge and it ends up teaching you a lot about the simplicity of the OS and beauty of OpenRC.
Otherwise, Debian is a very reliable and easy to use/install OS and well tested on Lenovo hardware. Failing that, install Ubuntu which has invested much in compatibility with ThinkPads, specifically X1's but those are expensive and cheaper models are often just as compatible.
If you have problems running wireless, the easiest solution is sometimes to figure out if your wifi card is officially supported and swap out the one in the PC with an older and more compatible wifi card.
Sleep and hibernate are often problematic. Find a workaround that works well enough.
You might be surprised how even people as aggressive or stubborn as this might still find it embarrassing or unreasonable to refuse a lunch-and-learn. Maybe sending a mass invite to a meeting on Python would be seen as passive aggressive, so just start socializing smaller ideas with key people. It sounds like you'd have to be cautious about upsetting people but I think it's always worth it to at least find the most polite and diplomatic ways to suggest things you want at opportune moments. The times I see this backfire is when people let it monopolize their attention or bring negative tone/attitudes.
Perhaps these devs prefer the "fix it ship it" mode and don't see the value of learning particular skills. I find it a balancing act to be pragmatic and it involves being skeptical about investing time learning new tech. With everything there is to possibly learn, you have to ignore almost all of it... until (at the latest) it's industry standard in the niche you specialize in and then you have to be open minded that it's worthwhile to the company and personally.
I have found in cultures like you describe "decision matrices" can be helpful because it allows people to consider costs, risks, do preliminary investigation, etc. That is a sort of way of providing encouragement and permission to learn things and innovate. Lunch-and-learns are another tactic to force people (or give them an excuse) to learn. Neither of those should need managers' approval to do (it's just creating a meeting invite for lunch or a wiki page). If these don't work the problem is probably the seniors internalizing management's whims instead of pushing back. But the engineers won't push back if they've never taken time to learn.
But if you asked it to generate a story in the style of Stephen King, whose works are still copyrighted, then does King have a copyright claim even though he didn't operate the AI?
Upgrades can be fun for end-users, like anticipating a new music album. And after feature complete there are features that can be added without muddying a product, e.g., adding plugin support, skins, additional language support, new platforms, performance improvements. And at a certain point in time it's reasonable to no longer make bug fixes in legacy versions.
Building a moat? If we're going to have AI at least it's more democratic to have the tools widely available than only accessible to governments and corporations. The basis of OpenAI's worry is explicitly that average people will be able to train their own LLMs.
This is my experience with good managers but also that they sympathize with the difficulties and successes of design and implementation otherwise they'll act in a way the hurts morale. Here it helps being a former individual contributor in the role being managed.
The mainstream culture has been pretty adamant about stamping out counter cultures, so it's merely hidden. I think the algorithms and filters are not blindly homogenizing culture but are rather used as tools by humans knowingly to connect to and persuade each other.
I was pretty impressed when I started giving it pseudo-code that it converted to Java code and could execute, and that it can function as a full Linux terminal and imagine files that would exist on a fictional computer, letting me cat file contents, ping servers, SSH (as long as you reassure it that it's purely hypothetical)
Crypto/payment startups are a sort of sudoku distracting VC's and software engineers, building an abstract world of endless technical complexity that passes for fulfilling work.
Congratulations this is an amazing application of technology and truly innovative. This could be leveraged by a wide range of applications that I hope you'll capitalize on.