CraigRood

I don't see them capturing anything at this point. If inference was profitable then they could compete on price/model and capture the market. Then increase price and pay back the model training.

Feels like they are just pulling in as much as they can whilst competing on capabilities instead. At which point its a case of who can last the longest.

Doesn't feel like Uber/Netflix.

Schooling is more than just exams, I'm sorry. There is no need for a cell phone in a classroom.

Not really a comparison when the spend on YouTube was x10 smaller, and Googles core business has always been profitable beyond any hobby spending on YouTube.

I'm FAR form an expert on this, but I believe that the operating costs such as power + cooling form a big part of the lifecycle. I have no doubt that at some point within the 6 years that are being booked, that replacing entire working racks won't be more cost efficient.

I use auto-complete mostly, so I'm somewhat relieved. When I do need to use the agent, I don't think I will use all of the tokens.

$10 a month for auto-complete on a good UX is good value IMHO.

That is actually nuts.... I'm trying to understand the true costs of AI, wonder how I plug this in!

I'm far from an expert here but isn't that spot price rather than future deliveries? Few people pay for actual spot pricing because it can go the other way, and you want known pricing. You would have a forward contract to delivery gas at say 20p. This is a known price for operation and likely has profit baked in anyway. The excess is what we see now. They can't just switch off as they have a contract to fulfill, but the grid doesn't need the excess, therefore priced at a negative.

Is this because at one point <username>@facebook.com was a valid communication method? Great concept to be fair, but once you pull back the first layer you can immediately see its problems.

In principle yes, but all metrics so far suggest they are losing money every user interaction. There is very little network effect with these tools so It's not like they can start cutting back on staff and feature deployment.

I don't 'like' Jira, but it gets the job done. It's so easy to onboard users and assign tasks/issues across orgs. Structure is fairly simply and the filters with subscriptions is powerful. Android app that I use on my work phone just works.

The growth is across the family of products (inc Instagram and WhatsApp) not Facebook itself. Facebook itself is a zombie, and I don't believe they have a way to innovate out of it. I'm not going to predict the end of Meta, they have more than enough products, but agreed that it's actually quite difficult to understand who's really left.

Windy.app looks good visually, but once you start using it, the UX is all over the place. Always find it frustrating.

It's less about paint a picture yourself, arguably there is little to no value there. OpenAI et al, sell the product of creating pictures in the style of their material. I see this as a direct competition to Studio Ghibli's right to produce their own material with their own IP.

I have a thought that whilst LLM providers can say "Sorry" - there is little incentive and it will expose the reality that they are not very accurate, nor can be properly measured. That said, there clearly are use cases where if the LLM can't a certain level of confidence it should refer to the user, rather than guessing.

I'm confused by your statement as the article suggests you can get a boarding pass via email.

I don't think users understand the risks. I'm broadly accepting of the protection of end users through mechanisms. Peoples entire lives are managed through these small devices. We need much better sandboxing to almost create a separate 'VM' for critical apps such as banking and messaging.

The whole notion of "Vibe Coding" was to accept the output regardless and prompt forward. Anything else is moving the goalposts. If you can't accept the outputs and you need an in-depth knowledge of code then these LLMs are not ready for this task.

Awesome post and fun read given I'm a PureGym member myself.

I 'got around' the PIN/QR Madness after 1 week by getting key fob. Now I don't have to ever open the app...

Attendance API looks to be worth playing with! Nice Bonus.

I run Gitea too - Seeing what is happening over at GitHub solidifies my decision.

Not too concerned over my public facing repos, Amazon and OpenAI seem to love 'em! I have the ultimate control over my private repos (nothing juicy). I can't say I trust Microsoft not to do something I don't like at any point in the future.

Edit: I should say I wish phabricator got more love, that was a great tool!

These tools are literally being marketed as AI, yet it presents false information as fact. 'using it wrong' can't be an argument here. I would rather then tool is honest about confidence levels and mechanisms to research further - then feed that fact back into 'AI' for the next step.

I was playing with it yesterday and every single session gave me factually incorrect information.

Speed and ease of use is one thing, but it shouldn't be at the cost of accuracy.

Thing is, even with users that don't use the quota, these AI companies are still losing money. This isn't a case of the small users paying for large. The true costs of AI are yet to unravel.

Part of my day job is Warehouse Automation - not Amazon!. I would agree with you on being slow, but it probably suffices to what Amazon want to achieve here. If your entire process, so stow, store and retrieve is automated, you wouldn't use these "pods". A lot of these problems seem so simple and easy to automate out, but it's really not!

ChatGPT et al are quite neat, but the interfaces are not great at all. For example, I'm going on a trip to Paris in a couple of months. I can ask a LLM what to do but it will just give me a list. There are no visuals, click throughs, maps, tips, experiences. blending 'AI' with search and a 'reader' can create a much better experience.

For what it's worth, I was able to add my custom domain to iCloud under this exact scenario without any issues. This was 3 years ago, so I don't know if anything has changed, and I didn't have the 'catch all' limitation either.

I've personally noticed loads of crawls too, but what has been interesting is they have been crawling effectively the same pages, over and over, for months. Also, they don't seem to be caching DNS at all.

I guess I'm in the minority that I don't care too much about crawlers, even the use of my work for AI, but I think the current hit rates just can't be justified.

I'm not sure how promising? I can't help but see how easy it would be to change API endpoints to a different platform.

I don't have any expertise here but my assumption would be the studios have a better way to manage digital content and rights compared to previous. It could very well be they have content available, free of rights, that can be uploaded to YouTube for monetisation. As others have mentioned, there are effectively no hosting or bandwidth costs associated.

From my limited experience working as systems for a large retailer. Thermal paper is more expensive, but the printers are cheaper and more reliable. It's not massive - I would estimate about ~20% difference - of course this can depend on print counts. Having printers down can have a much larger operational impact than the difference between types.

I miss Phabricator. Whilst it was certainly overkill for personal use, the additional applications meant I could do so much in one place and have everything integrated. Funny how what we have now, including GitHub feels a step back.