@FearandLoathingintheMotte's banner p

FearandLoathingintheMotte


				

				

				
0 followers   follows 2 users  
joined 2025 December 27 15:22:00 UTC

				

User ID: 4101

FearandLoathingintheMotte


				
				
				

				
0 followers   follows 2 users   joined 2025 December 27 15:22:00 UTC

					

No bio...


					

User ID: 4101

We can see the answer to this in every discussion on social security caps/contributions/clawbacks or British discussions on the "triple lock"

Hint, even if the welfare formula is mathematically destined to out compound the entire country's economy (triple lock) the voters will never, ever, EVER vote away their gibs.

Seconding this. I have a paid subscription to both openAI and Google and openAI's image gen is currently the better once since they released last

I anticipate Gemini Omni or whatever to eclipse openAI's current offering sometime this summer

or if they will have to keep doing research to stay competitive.

I find this part very funny. Because if we assume any lab who stops doing R&D will be out competed by a lab still spending like crazy on R&D, then we're implicitly agreeing their R&D spend is worth it, even if crazy.

I doubt any of their models have been stand alone profitable. The break even must be crazy.

There's no such thing as a frontier lab who can cut their exorbitant capex and coast on the margins from inference, as that's a one way road to getting cut-throat commoditized.

I find this part very funny. Because if we assume any lab who stops doing R&D will be out competed by a lab still spending like crazy on R&D, then we're implicitly agreeing their R&D spend is worth it, even if crazy.

Your subtext mostly seemed like sneering at these people "I have finance friends I talk to who still say their jobs can't be automated at all with AI" and AI at large, so yes it did cause everyone to respond poorly, because you came across poorly.

Some predictions:

  1. Your finance friends are idiots

  2. white collar work is going to get a huge step change in productivity, and many paradigms will shift across e-commerce and information work as a result. It's hard to fathom how, but agent-to-agent interactions will be levered in many ways. Aside from just making email jobs faster.

  3. What happens to white collar work as a result of this step change is anyone's guess. It depends on Jevons' Paradox, the elasticity of demand for white collar services, and the latent demand for white collar work in our society. Also somewhat rate limited by all our social systems, for example, if lawyers get 10x as productive and demand for legal services simply rises by 10x to meet it, our legal system implodes, so we have big changes ahead! Accounting is a good example. Excel made accountants more productive, but then as a society we chose to consume accounting that is WAY more complicated instead of having less accountants do the same amount of accounting.

  4. I was initially expecting the white collar labour market/productivity situation to start getting weird this year, but its almost 50% of the way through the year and $LARGE_MULTINATIONAL_FINANCIAL_INSTITUTION still hasn't given me Codex or Claude Cowork (genuinely embarrassing...) but my hodge-podge of skills and "open 30 chatGPT tabs to ghetto parallelize" is already changing my output materially.

  5. Once white collar workers have ample access to a Codex or Claude Cowork tier harness with ~Mythos tier base models, shit is going to get freaky for white collar workers unless there is a HUGE increase in the demand for information work.

  6. There is an absurd amount of productivity untapped in simply using the tools we have better, let alone the fact they are getting measurably better month over month. ChatGPT was noticeably shittier at using excel in early march 2026 than it is today (pre 5.4).

  7. The roll-out of 5 and 6 will take way longer than I think, because institutions are SO slow. And all the idiots like your friends (and you?) who refuse to embrace these tools cause diffusion to slow. But once the snowball starts, it'll be sink or swim for those not adopting quickly, and that will speed up roll-out, one way or another.

I have no predictions on robots/autonomous cars, but podcasters I trust keep saying robots are quite far away still, which is super lame.

I can't think of a single demographic group of humans who are considered "good drivers"

Similarly, I've never once heard anyone say "yeah the drivers in my $LOCAL_AREA are great! I love it! We all get along on the roads :)" and I've heard every version of the opposite, so I'm assuming everyone sucks everywhere.

Terrible driving and a fondness for alcohol is something that unites just about every culture and demographic.

got so fucking mad at some of the quality of the takes on AI

It's actually one of the best ways to bait me. I thought that now that I'm old and wise, I would stop taking bait, but they're just so wonderfully confidently wrong and I cannot resist. It makes me so tilted, but the "I told you so" as we stand in the breadlines is going to hit so nice.

It is an open secret that current models don't pay for themselves

Are you aware Anthropic and OpenAI both have gross margins in the range of 38-70% (depending on how you measure it).

R&D is eye-wateringly expensive, but inference is extremely profitable.

We continue this forums streak of awful takes on AI. Your bait sucks but I took it anyway.

claude opus 4.8, which is only 10% better than 4.5

We're already on Opus 4.7, which according to Artificial Analysis scores 15% higher than 4.5. Yes, yes, there are many reasons benchmarks are stupid and bad and wrong and misleading. But your prediction here betrays your ignorance.

AI solves maybe 10 niche, self-value giving math problems that is considered „impressive” by the kinds of people who put calories into math which never seem to come out of math

I don't know shit about high-tier math or how useful it is to society at large, but again, your ignorance is on full display here. I don't really know who Terence Tao is, but I don't think he's paid money to post on twitter, so the math he does must have some value.

AI is still completely useless at real scientific research and reasoning.

This is already wrong. Can you substantiate this in any way? What do you think of AlphaFold?

General coding ability stalls and Anthropic + OpenAI are looking for spiky improvements to models plus tooling ideas.

Finally! An actual prediction. I could see this happening, we shall see.

They begin to think about targeting non-coding white collar work like finance and spreadsheet work since the models are not getting much better at JavaScript, having used up all of the JavaScript data in the entire world.

You're again over your skis. I work at $LARGE_MULTINATIONAL_FINANCIAL_INSTITUTION and its revolutionizing how I work, despite IT making our roll-out as retarded and slow as possible.

having used up all of the JavaScript data in the entire world.

You clearly do not understand RLVR or post-training. Around half of training compute is used on these now, not just reading "all the javascript". Also your javascript quips are an attempt to downplay AI capability. They can program in every language. You can even invent your own programming language that isn't in the training data, and it can code using that too.

2029: Someone is arrested in the United States for plotting a terror attack with GPT.

Another real prediction, i agree with this but wouldn't be surprised to see it even sooner.

Frontier models begin to entshittify as they are increasingly jailed to make them safer

This has been happening this whole time, and especially since 4o

but they're getting really defensive about the new AI tutors, AI nursing assistants, AI portfolio advisors, AI middle manager helpers, AI emailers and spreadsheet workers, and so on.

This is literally already happening

Basically no improvements of relevance are made but Opus 5.0 drops along with GPT 6.0. The models are really only 10% better than GPT 5.5 at making JavaScript code but are 20% more expensive and slower

You think it will take THREE YEARS to get to GPT6 which is only a 10% lift on GPT5.5? This is a prediction, and a really bad one. I look forward to you being wrong in 6-12 months. Mythos, which already exists, is likely a 10% lift on GPT5.5.

Also GPT5.5 is almost 2x the cost of GPT5.4 so the price increases are already happening.

Some startups focus on making data collection tools that could be deployed in workplaces to monitor activities so than LLMs can be improved at non-software tasks.

This is already happening

Everything you say from 2030 onward is so deeply un-credible I'm stopping here. lol, lmao even.

I'm also blocked

We should start a run group

I just googled this and confirmed.

I thought you were doing a bit lol

This is phenomenal, thank you.

I think that there's only like a 5% chance that this is satire only matched by that fake Bay Area polycule invite from a month back

What is this?

Incredible worldview, thank you for sharing

The average office drone does SO much more today than they used to. What are you even talking about? Unless the bottleneck isn't tech like "the slide deck must be approved by the VP who has no time to review and approve the slide deck"

Accountants in the 1960s couldn't do shit compared to the accountants of 2026, and yet we have a fuckload of accountants in 2026? Why? Because we decided as a society that we wanted to consume significantly more complex accounting than we had in the 1960s.

If the SEC et al. decided to allow simple cash/accrual accounting and totally threw out IFRS/US GAAP we could crash accountant employment by like 80-90%. This would also be a disaster for efficient price discovery.

Hell the IASB is dead set on making accounting ever more complex and retarded with their fixation on PV-ing everything. Securing accountant jobs for future generations!

Echoing Corvos' question

I'm currently on the $200 OpenAI pro plan due to my thirst for tokens but I'm about to drop to the $100 plan and I'm barely even using 5.5 at this point as I'm largely having AI do non-coding activities for me now so I can get away with way cheaper models for "summarize all my to-do notes for today and file them" so having deepseek do my filesystem work would be nice (but I love the codex app)

There's also a world in which were using mythos^2 size models and the sheer compute to deliver 1 token, let alone 100 million, requires an absolute fuckload of GPUs.

There's also a world in which the inference-optimzied GPUs being worked on now take off bigly and are 1000x more efficient and we need one inference data center per continent.

Or anything in between!

Based and america-pilled

I'm not gonna lie dawg, I've read this entire exchange and I'm 1) incredibly confused and 2) completely unconvinced Magical is Darwin or adjacent to Darwin (also confused who that is, but assuming a troll)

You've accused them of being a bad faith argument shifter here to troll but produced 0 evidence of this despite claiming it exists...

Why is 2025-26 the only time in the last 40 years Iran was not developing a nuclear weapon?

Maybe this implies they spent the last 40 years also not developing a nuclear weapon? Given that 1940s tech was sufficient to make a nuke, maybe they thought "almost having a nuke" was valuable enough and stopped enriching roughly around there and then just made lots of noise about how "they were totally gonna finish enriching any day now"

You think lower class Americans are eating raisins? Or frankly, beans? If Americans ate fruits and cheap staples like beans in any measurable amount there'd be a lot less obesity and hand wringing about "food deserts"

I note that much of US Hispanic culture seems to emulate African American culture

I recently witnessed a tiktok of a Dominican heritage American (I use those terms because I imagine she was born here, not in the DR like her parents) going on a rant and she sounded like she was doing a bad AAVE impression for laughs but that's genuinely how she talks

Three things

  1. I work in financial services in a large city, in my yuppie social sphere, almost all the white collar women either get their nails done, or paint their own nails. Your impression that it's "trashy" is wildly out of date*, much like thos forums opinions on tattoos. To note however, all of them typically get tasteful pastel colors and the size of their nails is "mildly longer than a man keeps his, with an almond tip".

  2. I do not give a singular fuck about painted nails. I do not know a single man who gives a fuck about painted nails. It's pure inter-sex competition. I'm pretty sure every anxious striver yuppie woman plays the American Psycho business card scene in their head when she sees someone with "nicer" nails than her.

  3. *the gigantic fake nails that to my understanding are very popular amongst the African American community, are incredibly trashy. To my boy brain, I hate how deeply impractical they are. To my (limited) aesthetic taste, I also hate how garish they are.

Early in my career I had the great pleasure of being on-boarded by a lady with ~2 inch long nails. It was absolutely hilarious to watch her attempt to navigate a laptop trackpad and keyboard. She didn't even use a mouse! Psycho behavior.

What are we measuring "under the hood" and how is it being triggered

You can measure which parts of the weights get activated in response to inputs and map out what they mean. You can even mess with them to change outputs.

Anthropic releases like 300pg system cards with every model explaining things like this if you want to get deep