Missed this whole discussion today because I was here in California, doing a ton of productive work, using Fable.
I think you guys are working yourself into a lather about this topic while other people are quietly getting a ton of shit done with Anthropic's models.
Would be nice to have a distributed, independent AIs, each being trained their own way. Maybe it would have to be a really slow training process to keep costs low (years even?).
Do I understand this correctly: somebody at MSFT thought it would be a good idea to provide internal LLM with unfettered access to ALL of the GitHub code? “Just like SQL has”?
The difference is that (A) SQL is deterministic and (B) SQL implements internal access control (and how well that works).
Prompts from non-authenticated user should have no access to any private repositories. The real question is: can you trust MSFT GitHub with your code, now that “outsourced” engineers are supporting it?
I am using Opis 4.8 xhigh, in OMP.sh coding agent (full agent built on Pi), with Matt Pocock Skills installed.
I don’t see a particular bump in code quality from Fable 5. In fact, it feels less reliable to me than my current setup. No sure why I am not seeing what everybody else is seeing.
Perhaps OMP/Pi (head and shoulders better than Claude Code) + Matt Pocock Skills already encode all the agentic improvements Fable has?
Opus 4.8 xhigh is my daily driver for everything. I'd say Fable's edge is visible when designing for a complex problem with no obvious, idiomatic solutions. It is good at greenfield designs, and good at pointing out the pros and cons of hard design choices. I now do designs with Fable, and do implementations off those designs with Opus. Pretty happy. For a while I was using Fable for everything, but burned through a lot of real money for not much value, I think for coding its slow and not at all better than Opus.
I had the same sentiment. In my limited testing, it didn't perform any better than Opus at all. It wasn't a particularly challenging taskset either, mostly just "add this simple feature" with plenty of context and very clearly defined scope. It worked functionally but there were much better and simpler approaches available. For the cost, I don't see how Fable can ever be worth it.
I really hope that NVIDIA will push as hard as they can towards installed open-weights AI. Whether it's SOTA models on the Enterprise DGX installations, or regular models on people's desktops/laptops, they have every incentive to sell more hardware by doing so. And we benefit from having a non-controlled AI working for our benefit.
I do not think this is actually AI: currently, there is a narrative (gradually dying out) that AI will replace software engineers and you don't need CS/Software Engineering education as a result. It's the "leaders" who listen to this.
People still learn math, despite the calculator existing. Accounts still learn accounting, despite Excel and accounting software existing.
If/when it does change in 12-24 months, I think companies need to take a serious look at the people in these “leadership” positions. If the quality of their thinking on big things like this is that bad, and so easily swayed by marketing and hype, then they don’t seem qualified for the positions they’re in.
> People still learn math, despite the calculator existing. Accounts still learn accounting, despite Excel and accounting software existing.
They do, but you need far fewer or none of the original workers whose full-time job this sort of stuff was.
Raw math does not matter, but what you do with it. Similarly, you could earn a (modest) living knowing nothing but raw HTML, JavaScript and a bit of browser tech not too long ago. That is no longer possible.
Programming and software engineering will be devalued. These occupations won't disappear overnight, but you will see compensation and growth stagnate until equilibrium is reached again. Currently, supply outstrips demand, and I do think it is structural, not just hype.
I'm certainly not creative enough, but I currently do not see demand picking up sufficiently; Gen Z is bearish on social media, VR was a bust, blockchain was a bust, software has already penetrated almost all walks of live and lines of work. There is no next big thing (Internet, ...) on the horizon, to unlock the next order of magnitude of demand. There is certainly more work to do still, but it very suddenly does not require the same headcount, but something like 5%-30% less. Lots of the remaining work will be around integrating LLMs into existing software, which does not sound exciting either.
One annoying thing is how long it takes for things to sway back into equilibrium.
It's getting quite exhausting having to endure all these major events of the 21st century and their consequences - 9/11 and the Iraq war, the 2008 financial crisis, covid-19, and now AI.
But I guess it's better than all out war and conquest as was with most of human history.
What I agree with is that things will come back around in 12 to 24 months.
What I don't agree with is that I also consider this to be AI.
In fact, when you use AI, the stratification of input is very clear. In the end, even in software engineering, the quality of what AI produces depends heavily on how you prompt it. And there's no way around it—AI will inevitably do better than most people. It's pointless to say to an encyclopedia, 'I know more than you.' For a human to beat AI, the only way is to dig deeper into the latest technologies, but that's something only scholars who are up to date with cutting-edge academic trends can do. Most ordinary people won't be able to win against it.
However, I think software engineering will continue to exist. The reason is the stratification of input. In the end, software skills might become something like a subset selection technique for prompting within a specific domain.
I wonder about his marketing channels. If it was primarily SEO, that has taken a huge hit especially for programming related searches since AI answers showed up at the top of the SERP.
That's one way to see it. Can't we also imagine that more and more people now rely on AI rather than humans to learn programming (or more accurately learn vibe-coding)?
Your perspective is valid, but personally, I think humans will do better in this area.
The reason is simple: AI only gives you the information it deems necessary and semantically related to your prompt. But humans don't just dig into local details—they give you a broader map. That's what a curriculum does, in a sense.
In that regard, I think human education will shift toward cultivating macro-level insight and perspective.
p.s I'm really amazed that you're the author of the IPython Cookbook. I found that book incredibly useful. The fact that I was able to work as an assistant to data scientists was also based on what I learned from that book. I'm personally a fan of yours, so it's surprising to see you here
By the same logic, you would need to license Google and youtube access. Just because it's interactive, doesn't change the premise. But yes, that exactly the arguments Anthropic and OpenAI will use for the regulatory capture.
Kramnik ((ab)using his status as a former world champion) accused several other top level players to the point that one of them committed suicide.
Niemann allegedly cheated (In my personal opinion as a decently rated but certainly not elite chess player, he probably did cheat in this particular game, though proof was never given and probably never will) at a high level chess tournament and was ostracized by the chess community for a while. He has since rejoined and continues to play at a high level.
Hans Niemann most likely did not cheat in the famous game where he won against Magnus Carlsen in the Sinquefield cup. Or at least, there is no credible evidence for cheating, and Carlsen actually did not formally accuse him of cheating either.
Hans Niemann cheated in online matches when he was younger though
> There is no evidence of suicide, and he had multiple drugs in his system when he died.
Polyintoxication is nasty because after the fact, you can't tell if someone underestimated the effects of the drugs or knew about them and used it to guarantee death.
I have no idea why this continues to be a popular opinion among laypeople. Even at the time actual cheating experts using computer analysis determined it to be likely fair play. His accuracy was rather pedestrian, Magnus just had on off day and played terribly. There is no proposed mechanism for how he could even HAVE cheated. He has continued to perform at a similar level under much tighter security enforcement. The primary accusation seems to be that he's poor at explaining the reasoning behind his moves, but anyone who's ever watched a Hans stream can tell you that the man simply cannot speak.
Interesting. Maybe I am not getting something: it sounds like his cheating accusations cover some kind of an online chess tournament. Why does this even matter? Why would people care that much? Shouldn't they be focused on face-to-face world championships, etc...?
Online chess is taken seriously as an e-sport, especially at the top levels, and is played for cash prizes. Imagine if the best cs:go or competitive programmers were found to be cheating. They would be ostracized.
For the future of AI, we need to look elsewhere.
reply