skippyboxedhero

I think that identifies an issue that is going to cause a real problem for the US in the future. The society is deeply politicised and polarised to the extent that essentially inanimate objects are regarded as having deep political and social significance. When there is political change, it is going to sweep back in the other direction.

It also seems like people on all sides within the AI debate have been fanning those flames thinking is will work in the short-term...and it won't. Big tech played that game in many countries in the early 2010s and it didn't end well.

The use-cases for data science and other engineers are different. AI is not uniformly good at all kinds of development.

There is an issue with execs pushing it though. You have people at the top of the company with little to no idea how people work attempting to micromanage tool usage. It is as if you had a group of execs determining what IDEs people could use.

No-one is getting fired because of AI. The start of this year is the start of companies beginning to use AI. The reason layoffs are happening is because of the massive overhiring after Covid.

It is far easier in crypto. It is much more difficult in futures...as the case of a recent repeat Presidential candidate showed.

That will change when people realise crypto isn't anonymous...but that day isn't today.

It is either not being offered in depth by anyone market-maker (part of the answer given the relatively small revenue opportunity) or it is being offered by people who aren't sophisticated enough.

Bookmakers offer markets on events where someone can know the outcome. The difference is that they have tools to prevent adverse selection.

Prediction markets offer none of those protections so the market structure is going to end up being very different (which is already happening, revenue opportunity from politics isn't huge). There are other examples of this around latency arb, market is going to be very different.

Also, I will point out that most insiders are probably going to be losing money too. All that you ever read is the final outcome, you don't read the stuff that happens before. Politics is, generally, not a good market because the actual event is driven by decisions made by people. Election markets are fine but political event markets are not good, even if you have inside information.

There is also threadfin, which I found a bit more friendly than XTeVe.

Above system works pretty well but had trouble with encode/decode speed somewhere. Tried with N100 cpu and still had the same result...probably user error somewhere but none of the options seemed to work. No issues with UHF so kept using that.

China issued a stable coin about five years ago. It is used for all retail payments (I believe, small value, payments for govt employee salaries, etc). Somewhat bizarrely, it is significantly more privacy-protecting than payments in the West.

Quite funny to read comments from people asking what use is crypto. Can tell they have probably never left West Virgina.

Don't think it would be that useful for Iran though as they are already RMB earners, and RMB financial markets are still a bit questionable (there is depth, I don't think anyone knows why this depth exists or what it is actually for, just state-linked banks moving paper between themselves furiously).

My Anthropic rep told me it was just around the corner...you aren't saying he lied to me? Can't believe this, I thought he was my friend.

The assumption that the other person made was that I would only use it for coding. If you look through my other comments today, I suggest that they are useful for performing repetitive tasks i.e. checking lint on PR, etc. Also, can be used for throwaway code, very useful.

I don't think the issue is with the model, it is with the implication that AGI is just around the corner and that is what is required for AI to be useful...which is not accurate. The more grey area is with agentic coding but my opinion (one that I didn't always hold) is that these workflows are a complete waste of time. The problem is: if all this is true then how does the CTO justify spending $1m/month on Anthropic (I work somewhere where this has happened, OpenAI got the earlier contract then Cursor Teams was added, now they are adding Anthropic...within 72 hours of the rollout, it was pulled back from non-engineering teams). I think companies will ask why they need to pay Anthropic to do a job they were doing without Anthropic six months ago.

Also, the code is bad. This is something that is non-obvious to 95% of people who talk about AI online because they don't work in a team environment or manage legacy applications. If I interview somewhere and they are using agentic workflow, the codebase will be shit and the company will be unable to deliver. At most companies, the average developer is an idiot, giving them AI is like giving a monkey an AK-47 (I also say this as someone of middling competence, I have been the monkey with AK many times). You increase the ability to produce output without improving the ability to produce good output. That is the reality of coding in most jobs.

AI isn't good enough to replace a competent human, it is fast enough to make an incompetent human dangerous.

I think are fundamental issues with the story that Anthropic is selling. AGI is very close, we will definitely get there, it is also very dangerous...so Anthropic should be the only ones trusted with AGI.

If you look at recent changes in Opus behaviour and this model that is, apparently, amazingly powerful but even more unsafe...seems suspect.

That doesn't follow logically from what I said. You should ask your AI for help with this. You are in need of some artificial intelligence.

You're completely right.

Just checked my subscription start date for Anthropic. September 2023, I believe before they announced public launch.

Sorry kid.

Anyone who has used Opus recently can verify that their current model does all of these things quite competently.

GPT-2, o1, Opus...been here so many times. The reason they do this is because they know it works (and they seem to specifically employ credulous people who are prone to believe AGI is right around the corner). There haven't been significant innovations, the code generated is still not good but the hype cycle has to retrigger.

I remember when OpenAI created the first thinking model with o1 and there were all these breathless posts on here hyperventilating about how the model had to be kept secret, how dangerous it was, etc.

Fell for it again award. All thinking does is burn output tokens for accuracy, it is the AI getting high on its own supply, this isn't innovation but it was supposed to super AGI. Not serious.

I haven't noticed a huge difference with other models but I agree that is definitely a strength (and CC has better tooling for this). However, I do think there are practical limitations to agentic workflows because of the relatively poor output vs humans. You can generate lots of code, but most of it will be shit.

Agentic workflows do have a place in well-defined, structured tasks...but I don't think that is what most people are trying to do with it.

Yes, 100% agree. OpenHands has self-hosted, KiloCode and RooCode both have a cloud option. I don't think you are able to pass a session around with any of them. Codex seems to have comparable features afaik.

CC tool usage is also significantly ahead imo (doesn't negate the price but it is something). I have seen issues with heavy thinking models (like Minimax) and client implementations with poor tool usage (like Cline).

CC has had a period over the last six months of delivering significant value...but, of course, you can just use CC with OpenRouter.

I believe Centrica did some research before the Iran war and found that if we were able to get gas for free, energy bills wouldn't fall and would actually rise over the next few years (because of the mix towards structurally high cost supply).

It says something that the people running the monopoly cash machine are asking questions about bankrupting their customers/ability to pay but politicians are shutting their eyes and pounding onto the accelerator. What a world.

I was using web UI last night and it was unable to understand basic aspects of the task. Haven't seen it perform this badly since I began using two years ago.

Was trying to track token usage/index with Cursor, and was unable to understand that running `find` wouldn't show what was in Cursor index. Multiple times.

Anthropic models haven't been far ahead for a while. Quite a few months at least. Chinese models are roughly equal at 1/6th the cost. Minimax is roughly equal to Opus. Chinese providers also haven't had the issues with uptime and variable model quality. The gap with OpenAI also isn't huge and GLM is a noticeably more compliant model (unsurprisingly given the hubristic internal culture at Anthropic around safety).

CC is a better implementation and seems to be fairly economic with token usage. That is the really the only defining point and, I suspect, Anthropic are going to have a lot of trouble staying relevant with all the product issues.

They were far ahead for a brief period in November/December which is driving the hype cycle that now appears to be collapsing the company.

You have to test at least every month, things are moving quickly. Stepfun is releasing soon and seems to have an Opus-level model with more efficient architecture.

It has been a daily thing for 2-3 months.

classic :D i did think when i wrote that maybe AGI is already here, definitely worked with enough devs like that

I appreciate the work done here.

Been having this feeling that things have got worse recently but didn't think it could be model related.

The most frustrating aspect recently (I have learned and accepted that Claude produces bad code and probably always did, mea culpa) is the non-compliance. Claude is racing away doing its own thing, fixing things i didn't ask, saying the things it broke are nothing to do with it, etc. Quite unpleasant to work with.

The stuff about token consumption is also interesting. Minimax/Composer have this habit of extensive thinking and it is said to be their strength but it seems like that comes at a price of huge output token consumption. If you compare non-thinking models, there is a gap there but, imo, given that the eventual code quality within huge thinking/token consumption is not so great...it doesn't feel a huge gap.

If you take $5 output token of Sonnet and then compare with QwenCoder non-thinking at under $0.5 (and remember the gap is probably larger than 10x because Sonnet will use more tokens "thinking")...is the gap in code quality that large? Imo, not really.

Have been a subscriber since December 2024 but looking elsewhere now. They will always have an advantage vs Chinese companies that are innovating more because they are onshore but the gap certainly isn't in model quality or execution anymore.

Nope, just economic history.

People think the "new thing" is new...this has happened many times before. Macro tourists come in and think ten minutes with ChatGPT has turned them into an economist. Basic knowledge.

The correct distinction is: if you can't do something without the agent, then you can't do it.

The problem that the author describes is real. I have run into it hundreds of times now. I will know how to do something, I tell AI to do it, the AI does not actually know how to do it at a fundamental level and will create fake tests to prove that it is done, and you check the work and it is wrong.

You can describe to the AI to do X at a very high-level but if you don't know how to check the outcome then the AI isn't going to be useful.

The story about the cook is 100% right. McDonald's doesn't have "chefs", they have factory workers who assemble food. The argument with AI is that working in McDonald's means you are able to cook food as well as the best chef.

The issue with hiring is that companies won't be able to distinguish between AI-driven humans and people with knowledge until it is too late.

If you have knowledge and are using AI tools correctly (i.e. not trying to zero-shot work) then it is a huge multiplier. That the industry is moving towards agent-driven workflows indicates that the AI business is about selling fake expertise to the incompetent.

The Chinese were right about democracy being fundamentally corrupt.

To remove all regulation. Using economic policy to achieve social outcomes is insanity. Even the most free markets ultra would never suggest this, and you have people who are slightly to the left of communist suggesting this...it is worth asking why.

Switzerland and Italy. Two countries that are next to each other, very different story. Slovakia and Germany, adjacent (can't really use Austria because that is somewhere in the middle atm). At a high level, Eastern and Western Europe. Eastern Europe is thriving, Western Europe is struggling to decide whether to arrest people for committing crimes.

Not aware of what happened in the 70s then with Bretton Woods? Not aware of what happened pre-08? It has gone wrong multiple times before just for the US and the US exporting inflation is not good for either the US or the rest of the world.

As with NATO, I am not sure what the argument is...yes, other countries should take more of a role in their own defence...yes, other countries should take more of a role in their own economic governance/financial stability.

On this topic specifically though, I will also point out that whatever the faults of the US, no-one had a problem when Germany was exporting crippling levels of deflation (this is one of the main issues that wasn't solved after WW2 despite being identified by Keynes: deficit countries will be told they need to rebalance, but not surplus countries). Germany have solved that problem with catastrophically bad foreign policy destroying their economy (once again) but before this, they were the main source of global financial instability (China, to their credit, is probably one of the only countries ever to rebalance from surplus to deficit country...Germany has done this multiple times over the years and has blamed other countries every time).

Other countries using gold instead of US debt is good for both the US and those countries. Most of the "increase" has been due to gold rising in value and, fundamentally, it cannot solve the problem because the price cannot rise as much as it needs to but it should be part of a mix with SDRs and other securities.

To be clear too, this is not capitalism. This is corporatism. Large companies dictating economic companies is anti-innovation. It can only end with disaster and more control/corporatism because lower-productivity workers does not produce higher long-term growth. Temporarily you are able to get your bonus and stock options from the spread between imported and native workers but, eventually, demand and supply stop (and the US reached this point a while ago, which is why central bankers and politicians have had to intervene heavily to keep it going).

The end game for corporatism is shown in Europe where you can see a clear gap between countries that are built on non-zero sum systems which are thriving, everything just works...and then other countries which have been heavily corporatist for multiple decades, everything is collapsing, government function is both non-existent in many areas and reaching new highs of intervention into markets. Unfortunately, the Chinese were right.

By your logic, slavery was one of the finest economic policies. Cheap labour, how about free labour? Have we thought of that? Everything would just be free.

In the real world, the evidence is obvious: average productivity/wages drop, incentive to invest in labour-saving technology disappears, and you get multiple decades of stagnation. Every country which had unlimited, unfree labour has had decades of slow growth as a result.

Income growth in the working age population in the US since 1990 has been about the same as Japan, a country which is widely regarded as on the verge of economic collapse. US per capita income is probably 20-30% lower than it would be with first-order effects from immigration, likely much more with second order effects. Under any other circumstances with economic policy elsewhere, the US economy would be growing 7%/year now (and ofc, the answer for Japan's ills is apparently, you guessed, lots of immigration).

China is seeing secular reductions in production costs because of capital investment, not low wages. The peculiarly statist notion of American capitalists that the route to economic supremacy was large numbers of illiterate Guatemalans should go down as not only an economic failure but a moral one (equally of H1B).

Yes, also doesn't work for me. If the changes are simple, it is fine but if the changes are complex and there isn't a clear guideline then there is no AI that is good enough or even close to it. Gives you a few days of feeling productive and then weeks of trying to tidy up the mess.

Also, I have noticed, strangely, that Claude is noticeably less compliant than GPT. If you ask a question, it will answer and then try to immediately make changes (which may not be related). If you say something isn't working, it will challenge you and it was tested (it wasn't). For a company that is seems to focus so much on ethics, they have produced an LLM that displays a clear disregard for users (perhaps that isn't a surprise). Either way, it is a very bad model for "agent swarm" style coding. I have been through this extensively but it will write bad code that doesn't work in a subtle way, it will tell that it works and that the issues relate to the way you are using the program, and then it will do the same thing five minutes later.

The tooling in this area is very good. The problem is that the AI cannot be trusted to write complex code. Imo, the future is something like Cerbaras Code that offers a speed up for single-threaded work. In most cases, I am just being lazy...I know what I want to write, I don't need the AI to do it, and I am seeing that I am faster if I just single-thread it.

Only counterpoint to this is that swarms are good for long-running admin, housekeeping, etc. Nowhere near what has been promised but not terrible.

You can't write tests if you don't know what you are looking for.