“Do I belong in tech anymore?” from Ky Decker (not on Masto) https://ky.fyi/posts/ai-burnout
23 replies
@zachleat @khalidabuhakmeh Good piece. I have very similar thoughts and feelings.
. This is a management failure, management should not just push AI usage forward for the shake of AI usage but should also (together with the team) create strategies about what we want AI to do and what we don't want AI to do. Just because AI _can_ do something does not mean we want to delegate it to it. We have to discuss what we want our jobs to be. "You make meaning through the doing." 💯
@zachleat @khalidabuhakmeh
. How do we raise concerns about AI risks when the answer is always "more AI", "AI will fix this"? Merging bugs, merging 12000lines is OK, AI will fix it. Too much code to read? AI will summarize it. Merge conflicts skyrocketing? AI will fix.I feel a lot like the author on most examples
@zachleat @khalidabuhakmeh I wonder if GenAI isn't the nuclear weapons of our industry. We can build it and we can use it, but we may soon find out we shouldn't.
@ThibaultDu @zachleat I think most folks using AI have artifact derangement syndrome.
Focusing on the outputs but losing the intangibles that get you better outcomes.
Sure, AI can give you a result, but it robs you of what would have otherwise been an educational and intrinsically worthwhile journey.
It's a Faustian deal every time. For some, it's worth it, for others not so much. I think folks need to grapple with that.
@ThibaultDu @zachleat the other insidious part of it is we've all been trained to put our identities into our work. It gives us purpose.
For some, they see AI as a means of accelerating their worth.
Others see AI output for what it is, artifacts. Material in a Material world. It's hollow "content", and adds nothing to their feelings of self-worth or identity.
My concern is for folks building an identity on shaky premises.
@khalidabuhakmeh @ThibaultDu “but it robs you of what would have otherwise been an educational and intrinsically worthwhile journey” resonates so well for me
@zachleat @ThibaultDu AI by design is meant to isolate us all from each other. I think that's the part I'm sad about.
It's a tragedy of the commons when everyone has their own private car to work without those pesky social interactions.
@khalidabuhakmeh @zachleat That's one of the strongest concerns I have raised at my workplace. If you are using your AI agent to discuss architectural design decisions that you would have otherwise discussed with your colleagues, that's a loss. There is value in friction, like they write in the article too. Would "you" rather end up working alone with your agents (and being hypothetically 3x more efficient), or same efficiency but continue have people around you?
@khalidabuhakmeh @zachleat I am glad we are starting to have these discussions, than you both 🙏
@ThibaultDu @zachleat that's not even accurate.
You'll never be as productive as you could be without the help of others.
We all have blind-spots, weaknesses, and misunderstandings.
Without additional perspectives you could spend hours, days, months going down a pathway paved of your own ignorance.
@khalidabuhakmeh @zachleat I agree 100%. I just chose to work under the hypothesis that GenAI _does_ yield efficiency gains to ask "even then, is that how you want to work?". But I agree that in the long run there are many issues with sustaining this way of working.
@khalidabuhakmeh @zachleat I use to say: "Show me usages where GenAI improves team cooperation. Show me usages where GenAI successfully maintains a live system for a year."
@zachleat @khalidabuhakmeh 100%. Well put.
@zachleat this shit brings out the worst of cognitive biases where the one time it worked makes up for the fallout of the 20 times it didn't. the mere potential to get something done faster seems to justify whatever cost.
and besides the other 200 times we used an AI tool nothing very bad happened immediately after! so it was probably a positive! /sI don’t see the situation as so bad (perhaps my employer is just more sane), but this https://www.wired.com/story/a-new-google-funded-data-center-will-be-powered-by-a-massive-gas-plant/ leads to deep depression.
Oof. I empathize with this person from my own experience burning out of tech 3 years ago and sympathize with how AI-accelerated enshittification amplifies those feelings.
h/t @zachleat
@zachleat I think this has always been an industry that is difficult to navigate if you care about building high quality things that are not doing evil. Arbitrary deadlines with pressure to ignore security and correctness, dark patterns, selling out user privacy. The companies that pretended to care about this stuff usually actually didn’t and just needed the mask to attract talented engineers. (1/2)
Now that the masks are off, I know there will be a lot fewer jobs for me but reasonable use of LLMs is a very good signal now that a company is worth considering. (2/2)
@zachleat I've been thinking the same thing lately. I'm wondering if my current career has hit the wall, and it's time for a new one. This was my third career, so I would actually enjoy doing and learning a job I've never done before.
But with the USA about to tank the global economy, maybe I should not risk a new job? (As old as I am, you'd think I wouldn't have to worry about that. But I do.)