AutoGPT will take over your tasks and then your livelihood

Originally published at: AutoGPT will take over your tasks and then your livelihood | Boing Boing

Yeah well, it looks like it can’t even type in a font big enough for me to read on my phone so,

Forget Will Smith GIF


I’ve had several higher ups ask if I could replace staff with “the GP thing” or similar since it’s made the news. They have absolutely zero idea how it works, how challenging it would be make it applicable (in its current state) to the job at hand, but they’re seeing “oooh, so computer is cheaper than human…”

I look forward to at least one company having a very public blowup because they replace necessary humans with these proto tools in the interest of profits. And of course the C-suite will sail away on golden parachutes.


Don’t the CEOs asking about implementing AI to replace workers realize it would mean the end of capitalism? Without customers to buy goods, there is no economy. I’m all up for the death of capitalism; but I don’t trust the CEOs to manage it in a humane manner.


Yeah well, that’s down the road awhile, you see. Meanwhile, I’m gonna grab all the cash I can! And the stonkholders want some quick loot too-- they’ll fire me if I don’t make that happen!

(Same blinkered self-interest with issues like climate change, of course.)


Good point. Give a CEO a fish, and he’ll eat for a day; teach a CEO to fish, and he’ll destroy the environment.


I think LLM’s are going to be like autonomous cars, and never live up to the hype. Many companies have spent billions of dollars to develop self-driving technology over the last decade, and we are still no where near close to having a car you can get in, tell it your destination, take a nap and wake up when the car arrives there. The real world has too many variables.

I work in IT, and I frequently google PowerShell code samples, and I always have to adapt them to my corporate environment. I don’t see a day when an LLM can come up with code that will run in my environment without some kind of human editing. And that is pretty much what I’m doing right now with Google.




You may be able to automate parts of my job, but never all of it. In a bakery the size I work in, none of it. The machines are too big, too costly and don’t do enough. Software that can bake a pie will be a long time coming.


Looking forward to when AI can tell the car repair insurance callers that my car is too old to cover so they stop calling me.

I also need an ai robot to pick up the socks and charging cables laying around my house before my Roomba sucks it up and gets stuck.


Any Sci Fi that has had AI in it had it there as a warning of humanity’s hubris.




Good luck when there’s no similar problem to crib from in its data set. Plus there’s the risk that as the question gets more specific, and the number of relevant samples goes down, the more chance it’s going to bleed from one identifiable source with an incompatible copyright or license.


It’s funny, reminds me of the off-shoring of manufacturing to China. I was working IT at a company that did point of purchase displays and other marketing/salesy furniture. They decided to get rid of their manufacturing in Mount Vernon (just north of the Bronx) and move it to China… and keep only the design work and sales in the US - that was their “special sauce”… I told my coworker they will be shocked, shocked, when the company in China they’ve trained to do this gets its own US sales reps and dismantles them… which is ultimately what happened. So this is a case of “I’ll train someone else’s AI to do everything my company does”. This is going to end so well.


google still can’t figure out what zipcode i live in with my weather app half the time, and netflix keeps telling me to watch adam sandler movies. ( :scream_cat: )

at the same time i could see how there might be smart ai and dumb ai running around at the same time… someday. maybe.

with crypto and nfts seeming to bottom out, with driverless cars and the “metaverse” not panning out, it makes sense there’s got to be some new big craze for vc to dump their money into.

i keep waiting for uber or these gig things to die. are they actually making money at this point? ( other than say airbnb maybe? )


Yep. We had a software “neural net” model in 1995 to predict locations for gas stations - we’d collect the data on all existing stations, then have the neural net try to predict the best factors for new stations with high gas sales volume. Basically, the real answer was always “be first on a heavily travelled road” but if we trained a dataset with only say 1000 stations (a very very small dataset compared to a credit card company) like metro San Juan, we’d get outliers that the model couldn’t understand how to predict… and the most important outliers were the gas stations selling 2-3 times the average. So we’d predict the dirtier the gas station the more money it would make. Because the really busy stations had a lot of traffic and were pretty dirty.

I’ve seen this effect in engineering also - trying to look for things like transformer failures based on dissolved gas products in oil, and until you get to tens of thousands of cases you can’t really do very much… we just don’t have enough transformers failing to get the right dataset for what the leadup to failure actually looks like.

OTOH, if they run dashcams in all of the near autonomous vehicles for long enough they’ll get tens of thousands of car crashes… so maybe it works in 2040.


Twitter Account Suspended

I suspect it started to look to much like the Space Karen Musk.

1 Like

I’m still assuming that they’ll try to preserve it using a neoliberal UBI with this particular outcome to their greedpig game of musical chairs (which the’ve already won). What’s definitely not on the agenda for them is Fully Automated Luxury Communism.

Also, watch how quickly they move (again) from whinging that “people are lazy and don’t want to work” to "we just have accept that a whole swath of careers and jobs will be automated away permanently ".


Exactly this. That’s a next-quarter problem, not a this-quarter problem.


Don’t executives realize that these LLMs are great at sifting through vast amounts of data and keying on big trends and best practices for next to nothing in cost (at scale)? In other word their job at a fraction of the cost and the results would heavily favor paying the average worker more and executives less. There is a lot of talk about replacing the average job but not much about replacing positions with massive pay and risk.