Opus 4.6 is AGI in my book. They won’t admit it, but it’s absolutely true. It shows initiative in not only getting things right but also adding improvements that the original prompt didn't request that match the goals of the job.
Not even close. There are still tons of architectural design issues that I'd find it completely useless at, tons of subtle issues it won't notice.
I never run agents by themselves; every single edit they do is approved by me. And, I've lost track of the innumerable times I've had to step in and redirect them (including Opus) to an objectively better approach. I probably should keep a log of all that, for the sake of posterity.
I'll grant you that for basic implementation of a detailed and well-specced design, it is capable.
I don’t know if Opus is AGI but on a broader note, that’s how we will get AGI. Not some consciousness like people are expecting. It’s just going to be chatbot that’s very hard to stump and starts making actual scientific breakthroughs and solving long standing problems.
I'll be more likely to agree with anything being AGI if it doesn't have such obvious and common brittleness. These LLMs all go off the rails when the context window gets large. Their context is also easy to "poison", and so it's better to rollback conversations that went bad rather than trying to steer them back to the light.
There's probably more examples, but to me AGI must move beyond the above issues. Though frankly context window might just be a symptom of poor harness than anything, still - it illustrates my general issue with them being considered AGI as it stands today.
Claude 4.6 is getting crazy good though, i'll give you that.
This would be wonderful if it is accurate - instead of guesstimating, let people report their actual findings. I can confirm GLM 4.7 is possible on M1 Max and it can do nice comprehensive answers (albeit at 12 min an answer) locally. You can also easily do Mistral7B and OSS 20B and others. Structure it as a way to report accruals, similarly to Levels.xyz for salaries, instead of guestimating.
To refer back to the trite business book section - making something new is a Blue Ocean Strategy approach. Fighting for existing market share is a bloodied Red Ocean approach that Thiel called “competition is for losers”. So both benevolent and be greedy approaches recommend the same. Make a new puddle for everyone to swim in and you can focus on empathy instead of defense.
I’ve had several senior neighbours who passed away, after decades of living alone. We’ve always helped each other in one way or another and when they passed, in all cases I thought back about the last time we talked. In all cases the conversations remembered involved kindness. Either from then to us or from us to them and them being grateful. It’s what remains.
Giving kindness is most satisfying. It makes the receiver happy, but it makes you happy as well in a wat that kind of lasts. It’s an interaction that compounds on both sides. I think that’s why church groups, and mentorships, ans teaching, and advisory roles are satisfying. Teaching, giving, coaching all make life far more vibrant emotionally, and far less lonely.
Giving grows the pie, while zero sum games see it as affixed. Trying to compete in the latter will make you lonely. Trying to grow the pie in any kind of local community might make you see things differently.
I'll counter this comment with "watch who you spend your time with". I picked up that nugget from a successful (and happy) guy in North Carolina who does development and has really turned around the small town he lives in with his strategy for revitalization.
He also said to make your bed every morning, which I have done for years. Highly recommend this as it is an act of self care.
Yes, kindness is great, but so is being direct and knowing what you want for yourself. There's literally zero point being kind to someone who is going to take your energy.
If you have community, being kind is a great strategy, but if you are alone then it can be hard (and somewhat risky) to try to go include yourself in a group, especially if you are hurting. The risk is predatory behaviors from others seeing your weakness.
Personalized ads enable personalized lying by advertisers. Politicians in the 2016 election would target voters for party with enraging content while the other got shadow posts that lied to them about their candidate in a way that would not be seen, to discourage them from voting (Source: Careless People book).
Need is valid. The site is showing mostly flood watch warnings - maybe cluster topics? Also don’t mess with the scroll bar - maybe the ads are doing it, but it froze and wouldn’t move down for a while.
Thank you -- yes, the non-signed in front page needs some work. There's a lot of flood warnings, but if you choose topics with an account it should be a better experience.
And thank you for flagging the scroll thing. I hadn't seen it, but will check.
If everything is local, why the subscription? That 150 is instant incentive for me to prompt my own on Claude and get a more personal outcome right away. Margin comes from a moat, and local LLM is the opposite of that, especially if you need internet to verify subscription for local use at any point.
Fun fact - some of it may be a subset of all data and with trimmed outlying points, so when you set some stop loss conditions they get tripped in the real world, but not by your dataset. Get data from my sources.
Rare earth minerals are often all over the place, they are just very messy to get to and that gets in the way of EU pollution regulations. China is not a sole producer - they are just cheap enough to make mining elsewhere not worth the hassle. That will change fast if they bottleneck the supply.
There's a big find in Norway, largest in Europa according to new estimates[1].
The updated estimate shows that the total rare earth oxide (TREO) content in the mapped resources (Indicated and Inferred) has increased from 8.8 million tonnes in 2024 to 15.9 million tonnes in 2026 – an increase of approximately 80 percent. For the first time, parts of the resource are also classified in the Indicated category, reflecting a higher degree of geological confidence
The WSP report further shows that the proportion of neodymium and praseodymium (NdPr) can be increased from approximately 17 to 19 percent of TREO. These REEs are regarded by the European Commission as the most critical raw materials in terms of supply risk and are important in the manufacturing of permanent magnets for EVs, green energy and defence.
There's no mine there yet though, and they haven't yet determined if it's economically viable. So yeah.
And once enough panels start nearing the end of their lifetime, it's likely that we should be able to recover nearly all the rare earth minerals from them with proper recycling. They don't actually get used up the way, say, fossil fuels do.
Solar panels contain negligible amounts of rare earths, compared to the amount used in wind / gas / steam turbines. They're also still used in oil & natural gas refining (though less than in the past).
Fossil fuel generators are most reliant on them, wind less so, solar barely at all.
Oh, I completely agree—but they're so frequently used as a gotcha for why the rise of solar is just trading one "foreign master" for another. "Oh no, solar panels rely on rare earth minerals, so that means you have to kowtow to China!!!"
And it's true that there is some in them, so it's good to have at least a long-term answer for how we deal with them.
> Oh, I completely agree—but they're so frequently used as a gotcha for why the rise of solar is just trading one "foreign master" for another. "Oh no, solar panels rely on rare earth minerals, so that means you have to kowtow to China!!!"
> And it's true that there is some in them, so it's good to have at least a long-term answer for how we deal with them.
It's the old saying about a man and fish and giving vs. teaching.
Solar panels bought now, at least the quality glass-glass kind, doesn't really go bad in a way that makes them depreciate at-all-quickly.
If in locations that are not themselves at a premium, so lower yield only matters if maintenance overhead per yield becomes so bad it's cheaper to replace& upgrade, they can be expected to stay there for 30~50 years depending on how fast they'll mechanically fall apart after their warranty expires (which is expected to be the duration until which most stay alive). I'd guess something like an agricultural east/west fence install would stay more towards 50y and get individual modules replaced when they break, as they're easy to get to unlike roof/wall installs and the like where they're hard to get to and given they are very low complexity in mounting system ("fence panel") there's little engineering complexity in retrofitting a plain new future panel of the same physical size and sufficiently similar voltage/current.
I grew up in a world that didn’t have secure corporate jobs as a thing. That’s most of the world outside the US. There, you could get secure government jobs (if you come from the right family or connections) or you had to learn how to build a business yourself. Anything else, paid barely livable wage if at all.
The way to survive it was to 1) move to a village/small town where you could have a garden for fruits, vegetables, corn, chickens, maybe a pig or two for winter. 2) Young people lived with their parents while the parents saved up to build/buy their children their own flat or house. Children whose parents saved up enough would often start a family after getting their own place. Those who didn’t, co-lived with parents.
The secure middle class corporate employment in the US is getting severely downgraded by AI. While there is talk of universal basic income the reality is that many many companies depend on the surplus that middle class families enjoy spending and without it, vast swaths of industries will get starved as well. The solution is to show people quickly how to hunt and gather and farm as makers instead of just employees. Figuring out what is needed, taking on a small corner of needs somewhere in meat space or online, and planting there. AI has been fantastic at helping even solo founders with that. They need to encourage a cornucopia of ideas and experimenters as early as k12. They need to set up more favorable conditions for handling the admin side as well.
If the US government does not encourage cornucopia of AI-powered small business entrepreneurship and lets monopolies squash that early, they will end up with far FAR worse conditions. Any monopoly who keeps pitching “universal basic income” while actively avoids paying taxes, will end up forced into more taxes.
Big tech needs to make room for people to build and grow businesses (looking at you, Apple, for copying every successful app with a native, and you Meta for eating every social competitor) or they will end up paying for everyone’s universal basic income and then some.
If this country wants to survive the AI era, it needs to remove the pink glasses of “secure corporate job” and teach people how to plant, hunt, and gather as independent players in the market really fast.
reply