This doesn’t really surprise me. Most company leaders don’t have a detailed view of day to day work, they couldn’t step in and do every employee’s job. What they are good at is creating a clear story and direction that brings people together around a shared goal. That’s what Sam has done, especially in how he’s sold that vision to investors and raised billions. You could say the same about leaders like Elon Musk or Steve Jobs. It’s not necessarily a perfect system, but it’s often how companies grow and attract funding. No, they are not the perfect humans. It's just how business works.
I’ve been thinking that we’ll end up with specialized language models for different tasks, all connected by a router that picks the best one for each job. Put together, that could feel like a “super LLM”, not true AGI, but close enough that it might seem like it to most people. Because of that, I don’t think Mythos is a bluff.
I don’t think LLMs will ever be truly sentient, but they’ll likely appear that way. It won’t be real AGI, just something more powerful than any single human. Think of it as the oracle at Delphi. I'm not shocked that there's a better model coming. I'm sure there will be more powerful models to come after it too.
The junior engineer role isn’t going away. It’s just evolving. It’ll likely shift toward things like code analysis, reviewing work, and managing projects with AI tools. It’s similar to when we moved from hand-coded HTML to visual UI builders: the html code might not look perfect(it's actually crappy) , but it gets the job done. Over time, we’ll get better code management tools that will make code easier for everyone to understand. It sucks for the glut of coders that will need to adapt. But that's not new. How many web sites are written in perl, now? Did perl developers disappear? Some, but most moved on to other languages. That’s just how things progress.
Soon we'll have more programming projects popping up. It will now be financially profitable to create smaller projects that were losers before which will create new jobs. Or much more bigger project that could never be built before. So, AI won't take all coding jobs, they will evolve.
I turned a hobby into a career. I wouldn't recommend it. You lose the love for the hobby and the career doesn't match the hobby since you're mostly doing what other people need to do, not what you want.
I’ll just throw in a counter example. I turn 50 in two weeks. I programmed for the love of it and eventually turned it into my career. Maybe it hampered my love, a little, maybe. But the job and the passion are still different. I still write software on nights and weekends. But, maybe I’m a weirdo.
Hopefully this career will still be here until I retire. If not, I’ll try to adapt, maybe to something more hands on.
This is very insightful and also aligns/overlaps with the rather unpopular don't follow your passion mindset. one can still find some work that's reasonably enjoyable even when it's not your hobby or passion; just that such work takes a while to find and might not last forever so you'll be back to square one.
True, focusing on salary will only guide you towards a job you will hate. If I had the opportunity to do it all again, I would work towards a caereer where I had more interaction with people rather than a computer screen. It's something you might want to think about.
I'm with him on the connection of heart disease and sugar. The amount of sugar we as a society ingest is excessive. It can't be healthy. But I can't get behind the idea of an industry wide statin conspiracy. There's a lot of here-say in this article. If all the studies are wrong then the for and against studies are wrong equally. How can we even trust anything?
This seems to me to be a case of "It's true because I say it's true"
Avoid suspiciously low-priced bargains, give it a good several days of solid memtesting (I used to use Memtest86 and Memtest86+, and Prime95 for good measure) and you should be fine.
I feel this is going to be a short war. The war will be settled long before the November elections. The way I see it, there's going to be a "deal" and the administration will call it a win, "THE BIGGEST WIN EVER!" But I'm an optimist at heart. Keep calm. Think hard before you make a move.
It's going to be a short time until we see the newest version of the "Mission Accomplished" banner, I'll give you that. But the length of the war depends more on the power groups that rise up in Iran's wreckage. We don't seem to be helping up any friendly group, so that's not giving me much hope.
Why do you feel this way? Did you also feel Iraq would be a short war? After all it was really no trouble gaining air superiority and replacing any nominal authority with our own.
I’m totally open to theories of how America and the world at large will recognize a new Iranian leadership, but as is the strategy seems to be “create a power vacuum and pray”
The administration has said that they don't want regime change so that leaves just about anything as a win for the US. I see a big push to destroy the nuclear weapons material and capabilities. Once that's done, the US is out. It's a hard goal but it's the shortest way out.
Well, the news is reporting that the US is going to deploy a third carrier to the Middle East. Will it really be a short war? Or will it take a long time to destroy the remaining elements of the regime on the ground?
No, I don't see it. Current COBOL is running and meeting the needs. Rewriting the code would be a massive venture but even if Claude is able to map and rewrite the code there is 0 chance it can be trusted without major and expensive testing. Additionally, dealing with the new hardware that will be needed would be a major point of pain. Claude will help but why break something that's working. The better path will be to do a new needs analysis and then use Claude as one of the tools to make the transition. But the biggest block is getting management to pay for new code when the old code is not broken plus there is a chance of making things worse by doing the rewrite. Why would management take that risk?
reply