I would’ve already reached my breaking point by then.
POV: You’re Claude
3 Likes
Your build failed because of a security test, let me write a script to bypass that test! Not let me fix that security issue 
if you don’t watch everything it does it really will just go wild.
3 Likes
This thread was one of the first things I thought of when this news broke. I was actually curious if this was actually going to happen, which, it clearly did
1 Like
Yep. Recently got a hault on using tavily agentic as I guess it’s Russian or something too so lots of mess. Seems every other day more messy than the last.
At least if open ai struggles financially they will be so tied to the federal government they will get bailed out instead of crashing the market
Or Trump will use some magical executive power that’s never been used this way and declare open ai a state owned entity or something insane cause I’m not getting surprised by the crazy that’s happening day to day anymore
3 Likes
Either way, terrifying. Either more AI (meaning less consumer hardware) or we get a new- well, I’m not gonna get political on here again, lest everything I say be wiped out (again).
2 Likes
Brooks’ Law makes AI is a worthless tool. Reading unfamiliar code takes several times longer than writing it yourself. I write 99% of the code I read, which means I can read it incredibly quickly and easily. This is a necessity because I’m one of those people who write demented template & meta code from the third circle of hell in obscure programming languages.
4 Likes
I have been reading some stories about AI wreaking havoc at amazon lately. Seems there was an all hands meeting about it that made the news. What is the story? Is AI still being forced on the employees?
I was at one of our all-hands and heard our leadership claim that AI was giving our employees 30-50% productivity gains. When I asked what metrics they were using to cite this; I was told to stop asking questions like this…
Conversely, I am hearing rumors of great strides being made as companies begin to scale-up their LLM infrastructure. The gains are mostly in ‘reasoning’ models. If so, this might be really big.
1 Like
There have been a lot of news stories about AI and my employer lately. I am not allowed to directly comment on any of them or I’ll be fired. 
There is probably some AI monitoring to make sure I don’t comment 
I must phrase things carefully and deliberately…. Many companies have now set up metrics monitoring for “manual interventions” on pipelines and other similar code and pipeline metrics where achieving the goals without ai is impossible. So much as a single manual merge is tracked and counted against someone working in such systems. AI token use is also commonly in corporate environments being monitored and tracked in a public dashboard made available to management. These metrics likely will impact the next annual review cycle.
4 Likes
Monitoring talking about AI, with AI. Yikes.
2 Likes
Big Brother has great health insurance and I’m sick a lot
3 Likes
‘Reasoning’ models don’t use less power, at best they might be power efficient. They’re just more OpenAI hypeware to sell to credulous people.
Everything the cloud AI industry does is a short-sighted money-grab. Maybe they’re helping some people today but nothing makes it worth them destroying people‘s lives for it.
This whole situation has made me reevaluate whether I think The Internet as a whole is really ethical or even a net social good, given its similar infrastructure and destructive capacity.
On the ethical side, I’m definitely scratching my head trying to figure out why it needs to take so many kW just to use a ‘modern’ (AKA laggy, resource hogging) webpage, or why environment-damaging data centres should be allowed to exist at all in light of the ethical side of this.
The ethical side is… well, just look at Facebook. It’s exactly not a beacon of ethics is it? Surveillance, unethical human experimentation and manipulation, acting as a vehicle for radicalisation. But wait, that’s most of what The Internet is now!
The Internet was conceived for communication, but nobody realised how much it would be commercialised. We’re still able to communicate on it, but the methods of communication that we had full control over (e.g. your own personal email server) are all but dead, and we’re beholden to whoever has the fat stacks to run a platform for us to communicate on. We went from a distributed system that runs in the bedrooms of a few enthusiasts, to an unfathomable Goliath that feeds on everything we do. We shouldn’t have leant so hard on The Internet before evaluating whether it was the right solution.
4 Likes
Actually one of the better things that I use AI for is summarizing unfamiliar code bases and helping me find the sections I care about. There have been several times lately where it did this part flawlessly but then I had to hand edit my changes, but overall was still faster than I could have done it independently.
2 Likes
3 Likes
3 Likes
Let me guess, you’re supposed to use more tokens not less? (Which is insane because those tokens cost money)
2 Likes
yep they put you on a stack ranking based on how many tokens you use
2 Likes
You might even receive digital pieces of flare for using more.
Why not set up a few automated background review tasks in order to use as many tokens as possible then? I mean, you don’t even have to read the output. It’s almost like burning oil when it can’t be pumped away quickly enough.
There are people who are doing this to improve their numbers.
As someone who has worked here for 16+ years I have seen many large scale pushes to do x or y and consistently the people who manipulated their metrics end up tossed out on their ears. Word gets out about the loophole, abuse of the loophole spreads. Mass firing event.
I’ve seen literal entire buildings of people fired for manipulating metrics. So you won’t catch me doing something like that lol. But there are people doing it.
3 Likes
Well, good for you! Do you at least use way more tokens than absolutely necessary to stay on top? 
1 Like