The 1%ers are in the vertical, but the question is vertical to where?
It needs to a potential field with practical, economical, "real life" attraction well. I.e, robotics, real economic efficiency gains, manufacturing novelties.
The worry is that the 1% is attracted towards a non-practical money hole. I.e: Token burn for the lols, sophisticated software systems that dont provide actual value outside of giving NVIDIA cash.
I'm normally not the cynical type, and I am appreciative of Karpathy's efforts to educate people on developing language models, but this is exactly how it reads.
Every once in a while, everyone needs to touch grass.
It is fine as it is to make a boatload of money. All they have to do is seed ads into it. Look at how much money social media brings in from that model of "free/useless product" as flypaper to bring in eyeballs for the real money which is ad revenue. Adsense 2.0.
Vertical on the curve of throwing money away for a share of the pie, but if AI gains are zero sum, then all parties will end up spending their profit on AI to fight each other, just like most companies spend more on marketing to acquire customers than to deliver their product.
It's rough, but at some point there needs to be an arbitrary line.
I think 21 is used is many cases because that's supposedly when the brain has completed most of it's growing or something about pre-frontal cortex being fully developed.
> E.g, a baseball players trains to catch high-speed balls and they dont do it by: "ball velocity 50mph, vector:[1,2,3], run move hand command now"
That is a NN that learns a skill.
But that is not an Analyst. If it were ballistics, then the answer to "how to parametrize the launch to reliably hit the target" excludes getting the result through natural skill.
The problem lies in the need to get "AI" facing "LLMs": the latter create a need for reliability, for "AI".
Speech is an endowment of both those who give educated guesses via developed skills and of those who return answers like Analysts, who check and compute. LLMs create a confusion between the two, and they will remain a problem until an ability to act as Analysts - strictly - will be implemented.
The internal process of the blackbox oracle determine the reliability of the output.
Two abilities are very different: learning trajectories through empirical training ("increasingly catching thousands of thrown balls"), and determining trajectories through computation (a rational thinker at work). The former is a finetuned parametrized engine (a «NN that learns a skill»), the latter is an Analyst. The former is fuzzy, the second deterministic.
In front of fuzzy LLMs, which use potentially misleading outputs - text ("has it guessed or has it thought?") - the urgency of warranties of reliable output gets evident.
So, that they «simulate cognition [only] well enough» (Intended wrote), and that there are «fundamental limitations of models ... purely text based» (Kooi wrote) raises the urgency to overcome the "fuzzy" and achieve the "deterministic" - it is not that we can stall on a «fundamentally not the right tool for that».
Inventing an oracle calls for urgent striving to overcome the original weakness.
I think the most important aspect of high salary is job demand.
Perhaps once there was a high demand for good SWE coming out of 4 years of $$ tuition uni.
Personal anecdote: Got a buddy who dropped out of Bible school, worked at a auto shop while getting his certs. After 2 years about any dealership would hire him.
He moves, lands a 100k+ job. 2 years later, switches to a dealership in San Jose, earning 150+.
6 months later, he gets an offer to become foreman at a dealership in gilroy with a $5/h raise + foreman benefits. How it works is techs are paid flat rate: Paid per job, not per hour. So if youre good, you can really rake in the money.
Its been wild to see his career arc in real time.
He's highly trained, very handy, Nd he's damn smart at debugging.
Basically doing CAN bus diagnostics with an oscope and logic analyzer.
He earns more than me who got a MS in ECE.
On the flip side, his job is physical and he works more hours than me.
I get to so R&D, think about cool math, and code. But in another life I'd love do be a tech.
All that said, there's a lot of demand for good auto techs.
> Basically doing CAN bus diagnostics with an oscope and logic analyzer.
I'm skeptical about this; I can run my 'scope on the CAN bus and decipher the output, but there's no point. I'll use the CAN bus diagnostic reader to determine which of the 27 or so modules are sending malformed messages.
And that, itself, is theoretical - a broken module or poor wiring is still not going to result in messages so fucked up that the diagnostic reader can't read them.
It's like pulling out the 'scope to diagnose your LAN issues; you can do that, but you'd be a moron not to use a network packet sniffer instead.
No, when all the tools fail, they buy back the car or you lemon law it. Story over. It would not happen, does not happen, cannot happen.
I agree with the below poster, this reeks, no one is doing o-scope level debugging on a production vehicle. It's what someone would say that has no idea how cars are designed, built, and serviced in reality.
Lab Scope / Meter
2-channel
Bandwidth 3MHz
Sample Rate: Up to 6 million samples/sec
Scale range 0-100mV to 0-400V
Sweep rate 50 microseconds to 20 seconds (5 min in GMM mode)
It’s slow but it’s an oscilloscope. It graphs electrical measurements over time. They sold CRT scopes on carts for years and years.
Maybe you should move the goalposts and demand FFT or higher bandwidth.
There's demand but for all but the most specialized stuff (maybe NVH), the work is brutal and the flag hour system sucks ass. Got a warranty job? Better be happy with half pay. Got something that's an absolute bitch to get apart or something goes wrong? Yeah you're still getting paid the same even if it's not your fault. There's a good reason Ford can't find enough auto techs. There's no way I would go back to earning money as a mechanic over something like HVAC or being a millwright.
Extremely frustrating that the real doctor was so reluctant to get an MRI.
Why do they do this crap? I'd really love to know.
Similar experiences with our own doctors. It's pulling teeth to get tests done. If you every bring outside information, it's shutdown.
I have this idea about prompt injecting their audio scribe with repeated words. Maybe some part of the healthcare monolith will respond to a appointment summary with:
"You no listen. You no listen. Escalate. Escalate. Why make me in danger. Concern. Concern."
Mhh I've tried one-shotting double buffered IO a few times and it just can't do it. Maybe it's a prompting issue, maybe LLM's have a hard time with thinking about parallel processes.
I'll create a simple "framework" of what I know works. After that's there the LLM is fantastic.
love love this theoretically but in practice, at least in this “phase 1 of AI craze, unknown how long it’ll last” that someone coming out of Uni with deep knowledge of ABCDs of SWE will be unemployable unless she/he is AI native with harnesses coming out of their ears
reply