Thread regarding Oracle Corp. layoffs

@k5 lol. Claude 3 Opus isn't even comparable to Fable 5. same with GPT 3.5 Turbo vs 5.6 Sol. Light years difference.

@at There is a rumor where I live that an unused tech campus will be repurposed as an AI data center and the locals are pushing back hard against it.

@hq "Think about the jump AI made in the last 2 years. Now imagine 2 years from now."

Same as ORCL stock?

@k5

Sure code completion works OK, but even that has been falling apart recently.

what the he-l are you talking about? all development already moved to Codex, it works wonders, if have some IC5s who set up repos and skills correctly.

Think about the jump AI made in the last 2 years. Now imagine 2 years from now.

@hq are you talking about the jump AI made into the tank?

@hq You've just proven you know nothing about AI. There has been very little monumental progress in the last 2 years. The only thing thats happened since ChatGPT was released is any already existing programmatic automation that ALREADY existed has been rebranded as "AI". Sure code completion works OK, but even that has been falling apart recently.

I feel bad for rank and file folks being destroyed by ignorant C suite lemmings that are being twisted up and bent over by the shiny haired AI consultants from Deloitte, E&Y, PwC and McKinsey.

Getting a coding agent to do the work you intend to itself is a great talent. Unless you are tech lead or a low level manager whose sanity was tested for years by untalented useless staff who could magically crack an interview and some how found a seat next to you and you were forced to get work done by them, I dont think any meaningful complex project can be done by AI agents. So handing out some AI quota doesnt do the job.

Yes, you are smarter than Larry, Elon, Zuck, Altman, Dario, and Sergey. IT budgets will NOT be fixed.. they will reallocate engineering payroll to IT.

Think about the jump AI made in the last 2 years. Now imagine 2 years from now.

AI is here to stay because the oligarchs want it. You can be in denial, but that's the only reason it is mainstream.

@be

Haha...by your esteemed leader, an 81 year old man in a living Guy Fawkes mask.

I think we will see wide spread adoption of local llms where companies keep their data, have fixed costs and can leverage it we were it works for them.

@b7 yep agreed, however we were told today they are watching all employee usage. Basically get on the rusty merry go round or get thrown off.

AI is like just an advanced search engine

@b0 Nobody has seen the real cost of AI yet, right now all the price shock is still on very highly subsidized AI token cost. The real cost is many factors higher than we see today, this is why all the AI companies are loosing tens of billions a year.

Saying it needs to get cheaper when we are nowhere near the real cost is a problem that can not be overstated.

Ai is here to stay, but is it profitable? Only time will tell. Because users expect the prices to go down eventually.

@ah there are two constraints:

• fundamental one: LLMs are inefficient by their nature, they are brute-force, meaning they scale linearly with energy consumption. You cannot scale energy, it's the most fundamental civilization-level constraint.
• data for training and retraining: we already scrapped all structured and curated data we carefully collected through Internet existence. Probably some legacy things like medical records or something is still no full consumed. Theoretically we can scale data collection through mass adoption of sensory input collection, but it's very low quality.

@ah This is actually the 3rd AI "hype cycle" in the last few decades, this illogical exuberance is not new.

I wouldn't call AI down for the count just yet. It is extremely early in its life cycle and we have already seen dramatic changes between generations of models. It is really just getting started. It has a long way to go with many areas identified for improvement but there are no signs that the technology is up against any constraints that will prevent it from improving. You may as well accept that it isn't going away any faster than the Internet, cell phones, cars, social media or anything else you can think of... it is here to stay. You can be in denial if you want but you would be better off to adapt and leverage it to improve your life. There is likely going to be a popping of the AI bubble but it won't disappear.