[steve_adams_86]

I didn’t see this at first, and I was fairly shaken by the potential impact on the world if their progress didn’t stop. A couple generations showed meaningful improvements, but now it seems like you’re probably correct. I’ve used these for years quite intensively to aid my work and while it’s a useful rubber duck, it doesn’t seem to yield much more beyond that. I worry a lot less about my career now. It really is a tool that creates more work for me rather than less.

[iman453]

Would this still hold true in your opinion if models like O3 become super cheap and bit better over time? I don't know much about the AI space, but as a vanilla backend dev also worry about the future :)

[foobarian]

I was helping a relative still in college with a project, and I was struck by how lackadaisical they are about cut-and-pasting huge chunks of code from chatgpt into whatever module they are building without thinking about why, or what it does, or where it fits, as long as it works. It doesn't help that it's all relatively same-looking Javascript so frontend or backend is kinda mixed together. The troubleshooting help I provided was basically untangling the mess by going from first principles and figuring out what goes where. I can tell you I did not feel threatened by the AI there at all, if anything I felt bad for the juniors and feeling like this is what we old people are going to end up having to support very soon.

[muixoozie]

I had a similar experience recently.

Not sure how accurate these numbers are but on https://openrouter.ai/ highest used "apps" basically can auto-accept generated code and apply it to the project. I was recently looking at top performers on https://www.swebench.com/ and noticed OpenHands basically does the same thing or similar. I think the trend is going to get much worse, and I don't think Moore's Law is going to save us from the resulting chaos.

[docmars]

AI slop-fixing consultants, at your service! There's hope for us veterans yet. ;)

[intended]

It will be kinda like modern furniture, vs “Old Human written code.”

So many people won’t be able to get paid for the quality of their work, but the people who buy the cheaper product will get what they pay for.

[root_axis]

Let's see how O3 pans out in practice before we start setting it as the standard for the future.

[varelse]

Mamba-ish models are the breakthrough to cheap inference if they pan out. Calling a dead-end already is just silly.

[sydd]

We know that OpenAI is verz good at least in one thing: generating hype. When Sora was announced everyone thought that this will be revolutionary. Look at how it looks like in production. Same when they started floating rumours that they have some AGI prototype in their labs.

They are the Tesla of the IT world, overpromise and under deliver.

[WhyOhWhyQ]

It's a brilliant marketing model. Humans are inherently highly interested in anything which could be a threat to their well-being. Everything they put out is a tacit promise that the viewer will soon be economically valueless.

[bilbyx]

I hope people will come to the realisation that we have created a good plagiarizer at best. The "intelligence" originates from the human beings who created the training data for these LLMs. The hype will die when reality hits.

[gom_jabbar]

Hype is very interesting. The concept of Hyperstition describes fictions that make themselves real. In this sense, hype is an essential part of capitalism:

"Capitalization is [...] indistinguishable from a commercialization of potentials, through which modern history is slanted (teleoplexically) in the direction of ever greater virtualization, operationalizing science fiction scenarios as integral components of production systems." [0]

"Within capitalist futures markets, the non-actual has effective currency. It is not an "imaginary" but an integral part of the virtual body of capital, an operationalized realization of the future." [1]

This corresponds to the idea that virtual is opposed to actual, not real.

[0] https://retrochronic.com/#teleoplexy-12

[1] https://retrochronic.com/#on-accelerate-2b

[seadan83]

Religion too. Those that are told a prophecy is to come, have a lot of incentive to fulfill that prophecy. Human belief systems are strange and interesting because (IMO) of the entanglement of beliefs with identity

[steve_adams_86]

Generally speaking, I think it would. I’m open to being wrong. I think there is a non-trivial amount of hype around O3, and while it would certainly be interesting if it was cheap, I don’t think it would address important issues that AI currently doesn’t seem to even begin to accommodate in its current capacity to recognize or utilize contexts.

For example, I have little to no expectation that it will handle software architecture well. Especially refactoring legacy code, where two enormous contexts need to be held in mind at once.

[hatefulmoron]

I'm really curious about something, and would love for an OpenAI subscriber to weigh in here.

What is the jump to O1 like, compared to GPT4/Claude 3.5? I distinctly remember the same (if not even greater) buzz around the announcement of O1, but I don't hear people singing its praises in practice these days.

[jazzyjackson]

I gave up interest in GPT4/Claude3.5 about 6 months ago as not very helpful, producing plausible but wrong code.

Have an o3-mini model available to me on the other hand I'm very impressed with its fast, succinct, correct answers while tooling around in zsh on my mac. what things are called, why they exist. why is macports installing db48 etc. It still fails to write simple bash one liners. (I wanted to pipe the output of ffmpeg to a column of --enabled-features and it just couldn't do it)

It's a very helpful rubber duck but still not going to suffice as an agent, but I think its worth a subscription. I wanted to do everything local and self hosted and briefly owned a $3000 mac studio to run llama3.3-70B but it was only as good as GPT4 and too slow to be useful so returned it. In that context even $200/m is relatively cheap.

[lom888]

I don't know how to code in any meaningful way. I work at a company where the bureaucracy is so thick that it is easier to use a web scraper to port a client's website blog than to just move the files over. GPT 4 couldn't write me a working scraper to do what I needed. o1 did it with minimal prodding. It then suggested and wrote me a ffmpeg front-end to handle certain repetitive tasks with client videos, again, with no problem. Gpt4 would often miss the mark and then write bad code when presented with such challenges

[2muchcoffeeman]

Try Claude. I get even better code results.

[te_chris]

O1 is fine.

[tempodox]

No degree of cheapness will be able to offset the “creates more work for me rather than less” part.

[trhway]

>I worry a lot less about my career now. It really is a tool that creates more work for me rather than less.

when i was a team/project leader the largest part of my work was talking to the reports on what needs to be implemented and how they are going to implement it and the current progress of the implementation, how to interface the stuff, what are the issues and how to approach the troubleshooting, what are the next steps, etc. with occasional looking into/reviewing the code - it looks to me what working with coding LLM would soon be quite similar to that.

[trod1234]

Many of the major harms of these things were neglected, and downplayed even to this day people don't recognize just how changed the world has become. The mere delusion that AI will replace work has been used to justify mass layoffs.

The persistence of indistinct ghost jobs that are generated by computer for pennies to flood and bind with prospective job seekers (similar to RNA interference), has resulted in severe brain drain in many fields. Worse, the fact these people have often been forced into poverty as a result will have a lasting impact. You might have planned for up to a year out of work pre-AI and had the financial resources, but now how long does it take? Conversion ratios for the first step have changed by two magnitudes or order (from x100 to x10,000). What are the odds of these people finding a job given their finite time, and requirements that are un-automatable for submission (nil). The media keeps claiming that everything is getting better, the stats say so (while neglecting the fact that the stats are being manipulated to the point of uselessness/fabricated), but you have 1/3 of welfare payouts now going to these people in California (in the US), just for basic food.

When you can't find work, you go where the work is abandoning the bad economic investment and choice you made regardless of how competent you were. It is a psychologically sticky decision. When there is no chance at finding work, you get desperate, and many desperate people turn to crime and unrest. This was foreseen by a number of very intelligent people many decades ago, and ignored following business as usual.

The mere demonstration that we are unable to react in time is what gave engineers such great pause to write about these things, as far back as in the 70s. Hysteresis is a lagging time problem where you can't react fast enough to avert catastrophic failure given chaotic conditions, leaving survival up to chance. Its the worst type of engineering problem with real consequences.

Given how western society is structured dependently on labor exchange, its a perfect weapon of chaos and debasement in the value of labor, that effectively destroys half of its underlying economic structure (factor markets). This forces sieving conditions of wealth that become spinodal, and eventually falter under their constraints and spiral into deflationary trends over time.

Business wins so much that they lose everything. Its quite a disadvantaged environment and the general trend is that everyone is ignoring the pink elephant. Actions (and inaction) have consequences. When people don't listen and take appropriate action, consequences get dire, it hits the fan.

[cgio]

I agree that we are often missing in our analyses the true materialised impact of expectations by focusing on the validity of said expectations instead. Organisations, even if not laying off, are pausing hiring plans with a conviction that AI will replace some of the workers. It then becomes a self fulfilling prophecy to some extent. It doesn’t matter if it can, what matters is if it will. And to assume that people won’t place a bet is futile, as everyone does, and even if it’s wrong the market will allocate the losses to the baseline.

[trod1234]

There are two aspects of your line of reasoning that don't jive for me.

People can choose to not place a bet by not participating in the economy, or tying physical assets to it. In other words, de-banking, off-grid farming, unemployment on welfare (not their money, printed guaranteed loss absorbed by the baseline).

The assumption that the market can always allocate the losses to the baseline has already been shown to be foundationally flawed. It depends upon whether the baseline can absorb the losses to keep the market going, not the other way around. Those who believe in MMT don't pay head to the fact that money printing has caused societies to fail many times in the distant past, in the ever quoted phrase, but it will be different this time.

When the economic engine stalls, so too does order and money-printing/debt issuance (without fractional reserve) drives this as a sieve (which we've seen over the past several decades in the form of bailout, marketshare concentration, and consolidation).

Central banks set reserve allocations to 0% in 2020, adopting a capital reserve, risk-weighted system based in fiat that is opaque and stock-market tied (Basel III modified). Value is subjective, and fiat may have store of value, right up until it doesn't.

Of particular note, societal order is required to produce enough food for 8bn globally, without order and its now brittle dependencies, we can only feed 4bn globally. Malthus has a lot to say about population dynamics in ecological overshoot.

TL;DR Half of all people die when modern chemical production (Haber-Bosch fertilizer) and other food dependencies (climate) fail.

AI drives chaos and disruption. Its like throwing a wrench into a gear system, maybe it will stall, maybe it gets thrown out (still slowing it), maybe it runs rough wearing it faster further degrading the system towards failure.

When the baseline cannot absorb the cost in terms of purchasing power, it absorbs the cost from the resulting chaos in lives.

Intelligent people pay attention to history because the outcomes that repeat in history occur as a result of dynamics that repeat, and in matters where lives or survival are on the line risk management shifts from permissive to restrictive (where the requirements of proof are flipped).

[cgio]

Thank you for the thoughtful answer. There is a certain amount of cynicism in my post, to match the cynicism of reality unfortunately. Your arguments may be valid, but who cares to rationally think and act when they can easily observe and react? A collapse akin to your description would disproportionately affect the people who don’t have the power to do either, they just accept and suffer. In history, do we have any example where that was not the case, except revolution? Even with revolutions the respite is only perceived and of the shuffle to reach the new decision structures.

[trod1234]

I'm in agreement that self-fulling dynamics occur regularly over longer time horizons, I viewed what you said as pragmatism rather than cynicism.

As for who cares to rationally think and act when they can easily observe and react?

The problem with the latter is that its a false choice. The latter simply isn't possible in any effective sense. Certain systems and dynamics become a hysteresis problem, where the indicator is lagging ahead of when it actually happens; by the time you see the indicator it can be perceived, but ultimately impossible to react to in time.

There are also simultaneous issues with rational thought being deprived broadly through induction of psychological stress using sophisticated mental coercion and torture (which isn't physical). Rational thought is the first thing to vanish, and these methods act like HIV does in cellular systems (i.e. destroying the memory of the immune system making it unable to act, instead it destroys perception blinding people).

For some reason these things remind me of the Tower of Babel story in the book of genesis. It makes god out to be the bad guy, when it seems far more likely that the dynamics became destructive. All of Humanity has psychological blindspots that can be used to manipulate them in collective and through unity. Pride often lends itself to delusion, and blindness. Destruction usually follows, and confusion occurs naturally when delusion breaks towards a witnessed reality (as survivors, where others kept dying).

It seems like the translation is off, where instead of god, they meant the inescapable forces of reality. Albeit this is getting a bit into the weeds, its an interesting perspective.

Getting back to things, the major difference today when comparing to history with regards to revolution, we are in extreme ecological overshoot (globally).

Breakdown of order translates to famine so severe that half globally die from starvation. To make matters worse, nearly every economic system on the planet is controlled indirectly by one nation through money printing, and those distortions created are chaotic (fundamentally it shares many characteristics as that of a n-body immeasurable astrophysics system that has limited visibility).

When these things happened in the past, they were largely in isolation, and outside the geographical affected areas, assistance could be leveraged for survival. This is no longer the case now. If these things collapses, it all happens to everyone at the same time. Not enough resources exist to resolve the failure, and there are no tools that would allow correcting the situation after the dynamics have passed a point of no return.

Thinking about these things rationally, preparing while we can (before it happens), is the only tool that might allow survival long term for a few. Its important that survivors know what happened and how it happened, or it will happen again given sufficient time, and that requires a foundation. Needless to say, we have many dark times ahead.

A line appears, the order wanes, the empire falls, and chaos reigns.

I do not envy those who would have to somehow live through chaos, where nuclear weapons might be used by the delusional or insane.

[cgio]

I enjoy your thinking and your use of analogy. We agree more than is evident, but you think on an horizon that eludes most, myself included unfortunately. As you say, the psychological torture of mere lifestyle survival overshadows the rational concern for true survival. To some extent, we live through chaos, but don’t have the wherewithal to accept it as such and cling to a normal that is increasingly not normal at all.

[__loam]

> The mere delusion that AI will replace work has been used to justify mass layoffs.

AI might be the excuse but the reason is the end of zero interest rates and blitzscaling along with resentment among business leadership that some members of the labor force were actually getting a good deal for once.

[trod1234]

You are mistaken in more ways than one.

You can't claim wage earners have received a good deal when they are unable to support themselves with basic necessities, let alone a wife and three children (required for risk managing 1 surviving to have children themselves). This is largely why we have a problem with birth rate today with the old crowding out all opportunities for the young.

The problem you mention doesn't really have to do with AI. It comes down to purchasing power in the economy, not wages, and business has shown over decades they will not or cannot be flexible when it comes to profit.

Additionally, money printing puts both parties at each others necks through debasement of the currency. When the currency debasement (inflation) exceeds profit legitimate business not tied to a money printer leaves the market (no competition is possible).

When the only entities left in some proposed market cooperate, the market isn't a market, its non-market socialism without the requirements for economic calculation. This fails.

Neither parties in my opinion are getting a reasonable deal. Whose to blame? The cohorts of people printing money from nothing that call themselves central bankers.