Beep boop - this is a robot. A new show has been posted to TWiT…
What are your thoughts about today’s show? We’d love to hear from you!
Beep boop - this is a robot. A new show has been posted to TWiT…
What are your thoughts about today’s show? We’d love to hear from you!
THAT was the new theme song? I mean I understand flutes are not everybody’s thing, but… Ah. The new times. Well. On to the show.
Phew. I almost flew off the rails in the first quarter of an hour. This will be a difficult listening. Some bold statements I noted: innovation and the market will fix everything, not trust politicians to make the world worse, excuses of special interests, controlling for fallout is so incredibly easy, AI as the better, more generous political institutions (government), it’s a social experiment, let everyone do whatever they want and in then end, people revert to their best interest, let’s be optimistic! All in 15 minutes. This is crowding the zone, AI-accellerationist style. You could pick every one of those tropes apart, but it would take much more time than uttering them. I think I’ve had enough of that drink from the fire-hose for a moment. Too much of that, lately.
RIP fun, meandering TWiG. (But it’ll be fun for others.)
That sounds very overwhelming.
I miss the flutes. This new theme doesn’t do it for me (maybe it’ll grow on me)
I was losing my mind the first 20 minutes. It’s one thing to be bold on the AI future, but to say out loud the stuff about collective action and no need for a society (what he was saying in so many words) is beyond being positive.
This is the scary podcast, I want my TWiG back!
Seriously though, I agree, the first 15 minutes or seemed to be ignoring all the side effects of AI and push on a break the world, because AI will heal it somehow.
It tells you about how off the rails it went when Mike Elgan is the voice of reason! (I like Mike, so I mean that in a nice way. He has been a big pusher of AI over the last couple of years and even he was backing away from what was being said at times.)
Not to mention the guy talking is from California, and went to one of the best PUBLIC universities in the world and says that people paying into a tax system that has had no returns. It’s wild the hero complex where you just have to believe not one person or thing helped in life.
I’m with Benito, AI won’t be good at replacing programmers for some time. I can see them helping, and good for simple, internal tasks.
But when it comes to professional code, I don’t think it will be able to produce robust code.
As Benito said, people can’t even tell programmers what they want and they have to fill in the gaps. That is hard work and requires a good understanding of the area that the customer works in and which way to prompt the customer for the right information.
Also, a lot of what the AIs have learnt on is sample code and work in progress. Does it know to always escape input and sanitise data, check for buffer overflows? Many programmers today can’t do that or ignore it, until it is too late and the AIs are learning from this.
An experienced programmer is still going to have to go over the code, at least for a while. We can’t trust the AIs to give us reliable results on subjects we know, how are non programmers supposed to be able to tell whether the code produced is safe for use in a commercial environment.
It would be good if it worked as well as Leo believes, but I think we are still a long way away from being able to blindly trust what an AI produces.
I’ve experienced interactions with ChatGPT where it makes very odd changes unrelated to what it was requested to do. This implies, to me, it doesn’t generally produce repeatable results, and it’s my understanding that the entire concept of “temperature” is basically all about how “creative” it gets. If I had a software development team member who appeared to act randomly, I think I would be asking to have them replaced with someone nice and stable, whose output is predictable and it fits in with the rest of the team’s expectations. It may get to the point where you can bring on an AI helper and train them like you would a new team member, but until that happens, I assume they’re better at “brain storming” that feeds through an actual human team member to bring their output into proper scope.
Regarding the EU AI restrictions, there is nothing new there, it is just applying existing restrictions on businesses and software to AI.
It is already illegal to monitor employees in the workplace, so the AI decree is just reinforcing existing policy. It is the same with all of the other points as far as I can see.
It is a “nothing to see here” moment.
Patience - we’ll have all points of view on.
Oh dear.
Insert Michael Scott “no! god no!” gif here.
That was dire. Maybe I’m just not the right audience for this, because I find AI deeply boring and I already have far more than I want of it in my podcast feed. Perhaps, like TWiS and ULS, it’ll be one of those shows I have to skip over. But the whole point of TWiG was that it was an unfocused, rambling conversation, a group of friends chatting about nothing much in particular, and somehow giving it an AI focus kills that for me. It didn’t really get to be fun until well into the second hour, when everybody forgot about the “focus” and started wandering off-topic.
I miss TWiG already, and the TWiG flutes - they were just like Leo: bright, cheerful, with moments of manic laughter.
Amen to that!
Albeit, with a few days of thinking about it: there’s always a chance for aTWiT show to, gladly, wander off-topic and meander just like good old TWiG. Especially once it’s the regular cast again and most of the topic simply seems kind of “been there done that”. It’s got to happen with a single topic themed show.
Granted: I have not commented much on shows lately. At least newTWiG was new and different enough to elicit a little bit of action and rebellion. Wandering and meandering is wonderful for the ears - like trodden-in, old, favourite pairs of shoes - but they are also not getting oneself out of the comfort zone. However, one might argue that I love / pay TWiG to be one of my comfort zones and not necessarily the opposite. I can see how this might get repetitive and boring for the hosts, thought. E.g., they really seemed to hate the change log in the end. I always kind of liked it: drums, everybody makes fun about how little goes on at Google, drums again. That was the weekly ritual. But every once in a while, one needs new rituals, I suppose… That, and, with that amount of money in and attention on the subject, the positioning might lend itself much more to pick up some ads.
I just came across an interesting article I wanted to share in this thread: Technological Singularity: Are We Approaching the Event Horizon? | by Ilya Ageev | Feb, 2025 | Medium . Very interesting framing of the current state of affairs on this topic being akin to passing an event horizon of change.
I do think the show will go back to what it was when the normal hosts are both there. Plus guessing the guests won’t be weekly. I will say, I love having smart people on even when I disagree with them. Just hope it’s not marketing type folk from inside the AI bubble. Yes it’s a bubble and we know this because Google and Microsoft have already moved on to the point where the AI features are included as just another feature.
I’m inclined to agree. And I do know I can enjoy listening to podcasts on subjects I’m not that interested in, if the quality of the conversation is good enough.
Podcasts are an essential part of my battle against tinnitus, providing a constant buzz of interesting conversation to drown out the phantom noises. When the radio show and ATTG went, I had to cover a deficit of 6 hours of content, which meant spreading the podcasting net wider. As part of that exercise, I added a podcast which is just the same three people each month talking for a couple of hours about the games technology space and the games they’ve played. Not a subject that interests me much, but their conversations are such fun that I always look forward to it. I just wish they’d stop apologising for running long, because that’s what I need from a podcast - wall-to-wall conversation.
But maybe not about AI.
I was never that big on the TWiG theme song but it fit with the motif of talking about the cloud.
This new song was hopefully written by AI. Otherwise what are we doing here!?
Going back to the role of AI in programming. Producing code that works is not the issue. Designing a system that successfully meets a business or functional need is. We used to call that systems analysis. I’ve been away from the field for some years so I don’t know what it’s called now. This is entirely a human problem. It takes considerable skill to elicit from people just how a business process functions, what all the edge cases are, how automation can help and what the future process should be. I don’t see a role for AI in any of this.
I think most software producers call it a waste of time, judging by the rubbish they kick out, even after detailed workshops.
I experienced that too, as far back as the 1980’s. Us programmers needed the work that the systems analysts did to capture the user requirements, but it was delivered in a format (decision tables) totally unsuited to the programming language we used, and with no understanding of good process flow, so it would have created unmaintainable spaghetti code if programmed from the analysis.
I always had to sit down and flowchart what they’d identified, from that identify the principal activities and key decision points that worked well with the programming language, then draw up a new flowchart which would produce easily read and maintained program code. Pattern recognition was a very important programming skill. I don’t know if I was right, but my programs often lasted a very long time, and some were extended to perform additional functions by later programmers, not rewritten. Keeping my fingers crossed that meant they were maintainable.