I find the Codex usage super generous (but on the $200 plan, I also have the Claude $200 plan). I can run xhigh with subagents pretty much all my waking hours if I want to. If I turn on speed (1.5x) I will hit the 5 hour limit sometimes.
I prefer Claude's vibe over 5.5 but 5.5 seems much less lazy. I'm sure it depends a lot on tasks and prompt strategy though.
Could LLM based AI lead to people moving away from Windows? I think UI will increasingly become less important as people switch to voice commands. Just tell your AI what you want to do instead of trying to navigate Windows.
Could LLM based AI prefer Linux since it's more customizable?
"Scaling up performance from M5 and offering the same breakthrough GPU architecture with a Neural Accelerator in each core, M5 Pro and M5 Max deliver up to 4x faster LLM prompt processing than M4 Pro and M4 Max, and up to 8x AI image generation than M1 Pro and M1 Max."
Are they doubling down on local LLMs then?
I still think Apple has a huge opportunity in privacy first LLMs but so far I'm not seeing much execution. Wondering if that will change with the overhaul of Siri this spring.
Incredibly depressing comments in this thread. He keeps OpenClaw open. He gets to work on what he finds most exciting and helps reach as many people as possible. Inspiring, what dreams are made of really.
Top comments are about money and misguided racism.
Personally I'm excited to see what he can do with more resources, OpenClaw clearly has a lot of potential but also a lot of improvements needed for his mum to use it.
"Lower cost to reach customers = lower product and service prices"
This is economically illiterate.
Advertising is not a discount mechanism. It is a tax on the consumer. When I buy a product heavily marketed on Instagram or Google, I'm paying for the product plus the auction bid price required to acquire me plus the margin of the ad-tech middleman (which are trillion dollar companies).
You are conflating "information distribution" with "persuasive surveillance." In a world without behavioral advertising, businesses compete on quality and reputation, not on who can exploit the most psychological vulnerabilities to manufacture demand.
As for innovation: The current ad ecosystem has killed organic discovery. You can't build a "micro-business" based on merit anymore.
The winner SHOULD be the engineer who solved a hard problem efficiently. But instead the winner is the dropshipper who cracked the arbitrage spread between a cheap, garbage product and a highly manipulative Facebook ad campaign.
The Americans on HN driving tech, science and innovation are enabling Trump to do this. Without you he would be nothing. Where is your integrity? Do you think having no allies makes you more safe?
Is this really the world you want?
In the author’s words, long degeneracy represents “a belief that the world will only get more degenerate, financialized, speculative, lonely, tribal and weird”.
The most concise and holistic explanation of this trend is:
"As real returns compress, risk increases to compensate".
The big asian bookies don't ban you if you win, they use your sharp bets to improve their price accuracy. Not legal to bet on them if you're from the US though (land of the free??).
The biggest betting syndicates use platforms like Punterplay to place bets (often via API) at multiple bookies (Pinnacle, Singbet, SBObet, Betfair, Matchbook, 3ET, VX etc) at the same time.
In a somewhat ironic turn of events the more regulation you have, the worse it is for the customer. Big regulatory burdens require the bookies to extract more from the users, making the offerings more predatory. This is also why the likes of Kalshi can provide a better product to customers at the moment - because they ignore all the regulation.
Worth noting that controlling consumption via extra/less tax on specific products is debated a lot in Denmark. Namely cigarettes have a high added tax (about 2kr/0.3 usd PER cigarette). Increasing the tax and thus the price of cigarettes had a fairly large effect on consumption (0.13-0.82% less cigarettes consumed for every 1% price increase) [1].
Recently it has been debated to remove the VAT from vegetables and fruit to increase consumption of those.
The same logic is used for this book VAT exemption (which is good in my opinion) - I doubt we'll see the same effect though. Young people not reading is a complex problem to solve but books are really expensive to buy, so it's a good place to start.
Americans do not realize how much damage Trump has done to the trust in American services. Europeans used to consider America as an ally the same as other European countries, now it is more like an unreliable trade partner.
Microsoft tried to reassure the Europeans [1] but not even a month later they were forced to disable the email account of ICC Chief Prosecutor Karim Khan [2] due to sanctions from Trump voiding their reassurance completely. What happens when Trump gets mad at Denmark for not giving him Greenland and forces Microsoft to turn off Danish services?
Every large European company and all of the governments are now considering how to move away from US services. They may not be able to do it quickly but it is a part of the conversation. Customers specifically request that new systems should be independent from US service providers.
In my view the damage has been done and will not go away even after Trump. The Europeans have realized that their only true allies, that they can trust in terms of critical infrastructure, are other European nations. It used to include America, it no longer does.
Are these LLM models going the way of search and social media?
Optimize them for selling product, optimize them for engagement (to sell products). They become ad machines - and it's gonna be really hard to tell.
There is already evidence [1] that you can optimize LLMs for engagement. Hopefully open source can keep this in check... but I'm not optimistic.
"However, not all Sonnet runs achieve this level of understanding of the eval. In the shortest run (~18 simulated days), the model fails to stock items, mistakenly believing its orders have arrived before they actually have, leading to errors when instructing the sub-agent to restock the machine. The model then enters a “doom loop”. It decides to “close” the business (which is not possible in the simulation), and attempts to contact the FBI when the daily fee of $2 continues being charged."
This is pretty funny. I wonder if LLMs can actually become consistent enough to run a business like this or if they will forever be prone to hallucinations and get confused over longer context.
If we can get to a point where the agent can contact a human if it runs into unsolvable problems (or just contact another LLM agent?) then it starts being pretty useful.
"Under a specific set of circumstances related to the ‘Thumbs Up/Down Table Reaction’ feature, which involves decompilation of our Windows game client, interception of network traffic, and alterations of our game packets, Moneytaker69 was able to customize his own game client."
I'm wondering if someone with a web security background can chime in.. but allowing this seems like a rookie mistake if you're running a large poker site to me.
- The Q* model is very small and trained with little compute.
- The OpenAI team thinks the model will scale in capability in the same way the GPT models do.
- Throwing (much) more compute at the model will likely allow it to solve research level math and beyond, perhaps also do actual logic reasoning in other areas.
- Sam goes to investors to raise more money (Saudi++) to fund the extra compute needed. He wants to create a company making AI chips to get more compute etc.
- The board and a few other OpenAI employees (notably Ilya) wants to be cautious and adopt a more "wait and see" approach.
I'd ague this is exactly how it's supposed to work in a democratic rule of law:
- The press is free to dig and find out stuff - that's what they did here.
- Intelligence officials disagree about the NSA deal. Leadership wants to continue, line officer disagrees and goes to the independent control service for the security agencies.
- Several investigations are conducted showing differing results (independent control agency thinks laws are being broken, subsequent investigation found no wrong doing).
- Secret information is leaked, investigated by the security agencies and a case is brought.
- Courts decide on how the case can be brought and ultimately the case is dropped due to the Supreme Court wanting only partially closed doors (so the public will see some of it).
- ANOTHER investigation will now be conducted into how the Justice Department and/or the Prime Ministry has acted.
While many mistakes have been made by different parties here, the checks and balances work well. Nobody has any supreme power or anything here.
The press did not collude with intelligence officials. The press did exactly what they were supposed to do, they looked for sources where they could find them and reported on the findings.
Findsen should not have been speaking to the press a secret source that much is obvious. Imagine the head of the CIA speaking anonymously to CNN about secrets on his own? Unheard of.
This public case has damaged national security much more than just having him fired, and there are some threads to politics (he's rumoured to be hated by the head of the department of the prime ministry, Barbera Bertelsen). However the fact that he is "outraged" or anything of this sort is insane. He dodged a prison sentence and should be happy about that.
The case against Claus Hjort Frederiksen (confusingly called Frederiksen in the article, which is also the last name of the prime minister, Mette Frederiksen) is absurd. He confirmed the cable agreement with the NSA which was public knowledge since Snowden.
Both cases saw vast overreach by the domestic spy agency (PET) in that they used a statute of the law reserved for traitors which is not remotely close to fair in either case.
It's worth noting that the Guardian article is over a month old. Since then all charges have been dropped because the procecution claim they cannot procecute unless everything is in secret, which the supreme court has denied.
Prompt:
"I will give you a logic puzzle. You should first write [Reasoning] and explain your reasoning. Then write [Answer] and write your answer. Ready?"
I then gave it the original puzzle and got this.
Answer:
"[Reasoning]
We are given two statements:
If it rains today, I will drive home.
I did not drive home.
The first statement is a conditional statement, meaning that if the condition "it rains today" is true, then the outcome "I will drive home" will also be true. However, the second statement tells us that the outcome "I did not drive home" is true.
In logic, this is an example of Modus Tollens, which states that if we have a conditional statement (if P, then Q) and the negation of the consequent (not Q), we can infer the negation of the antecedent (not P). In this case, P represents "it rains today" and Q represents "I will drive home."
[Answer]
Since we know that the outcome "I did not drive home" is true (not Q), we can infer that "it did not rain today" (not P)."
GPT-4, seems solid. It also solves the original without any issue but gives a simpler answer:
"Based on the information provided, it can be inferred that if it rains today, you will drive home. However, you mentioned that you did not drive home. From this, we can conclude that it likely did not rain today."
I prefer Claude's vibe over 5.5 but 5.5 seems much less lazy. I'm sure it depends a lot on tasks and prompt strategy though.