TLDR: The train has already left the station.It's now the same as a car or a smart phone, it has been integrated into our daily life.

Quoting: scaineMan, I can't believe we're still defending genAI. As I've pointed out in many other comments, the top reasons I hear for the "negative resentment" are, in no particular order:

1. Negative impact on environment, slap bang in the middle of a climate crisis.
5. Driving a nuclear age .....(Meta, Google and Microsoft have now all commissioned their own reactors)

True.

Quoting: scaine6a, 6b, 2, 3, 7. (social impact and allegations)

These things are not relevant or have overlaps with (game) development. Sure, they are valid in other contexts, bu not this. This is also the issue with these AI Witch Hunters who toss everything into one bowl and don't differentate.

Quoting: scaine4. Slows down development (even in cases where developers claimed it sped them up, evidence showed otherwise)
6c. Societal impact - genAI "slop" now devalues everything on the internet. When you see something cool, you think "meh, it's probably just AI shite". Or it actually IS shite, in which case, genAI is on a race to the bottom, since the next generation of genAI will be taught on today's internet - mistakes will be compounded, biases reinforced.

Not really, unless you do use LLMs only locally (unless you have enough hardware), using only free versions of public LLMs and LLMs which are known to have issues, once a project gains a certain size threshold.

If you are properly defining the project, the feature list and all its stuff surrounding it, you can do amazing things properly coded if you keep oversight, test, debug, review code, make appropriate changes and hold the assitant on a very short leash.

You still have to read the code, understand its workings, architecture and program flows, to make a well rounded project.

And yes, you pay for assistance enough, to achieve it, but you can really accelerate development tool-assisted. You also have to do your research and do decisions.

On the other side, if you just throw a LLM some rough paragraphs, that are not well defined, you will end wit slop (and projects that are at some point borked).

Quoting: scaine8. Hallucination (multiple cases of invented bullshit, including court filings, leading to lawyers being debarred).

regarding coding, this does not happen, and any bullshit code is quickly remedied once you have tested and reviewed it.

Aside that, for (game) development that is not relevant.

Quoting: scaine9. Obnoxious marketing (see MS especially).
10. Diverting investment away from targeted solution, and into a financial bubble (because #7).

Quoting: scaine11. All genAI engines are built on plagiarised work, for which the original authors/artists got no recognition, nor commission. Same with code - all code was scraped, regardless of license, and that code can be regurgitated in new, OR snippet form, by genAI, without recognition of that license.

Not really true, if that LLM has general programming and knowledge about standard algorithms. For example, if there is some undocumented RESTA API not documented anywhere on the web and you reverse that and give it a summary how the program flow is and works, at least the paid LLMs are able to integrate that in your application.

Free LLMs and locally run LLMs are way more prone to that and mostly try to throw dirt at the wall and try to be sneaky hard coding stuff or trying to implement shortcuts to just satisfy the requirement that "it works".

Quoting: scaine12. Impact on website scraping from multiple companies building genAI models. Wikipedia in particular has had to actively block enormous ranges to prevent the scraping from leading them into financial run. Again, can't be bothered to find the link, but there's a Wikimedia blog talking about it.

Most web sites already have taken active measures against that scraping and that is sadly a fact of life now.

Quoting: scaineAnyone offering the "it's just a tool" argument, is being deliberately obtuse. They're basically arguing that the ends absolutely justify the means, no matter the cost.

IT IS A TOOL, especially one that has to be used properly and given enough input data that is not wishy-washy.

Quoting: scaineAnd the cost is high. Big tech has absolutely no morals, and this is a race to the bottom, fueled by literally hundreds of billions of investment that could have have so much difference elsewhere.

Corporations always have been that, not just yet, but that is not a justification. Just be happy, that we aren't already in full Cyberpunk 2077 / Shadowrun / Neuromancer territory.

As above stated, the main problem with AI witch hunters is, that these want to turn back time unconditionally to a non-AI world, what would be about the 1970s if you call Weizenbaum's ELIZA an AI. Or what would be the point in time for you? Perhaps you should differentate between the various forms of AI and not toss everything into a barrel and blaze it aflame?

LLMs and AI are nowadays a fact of life, deal with it. It is the same with smart phones: if you are against smart phones (and maybe "dumb mobile phones"), you will have major disadvantages and inconveniences NOT using them. This is the same with LLMs, there are useful applications with them, but I myself only use them for project development and not for other things.