I think it's fair to say that both Google and ChatGPT are roughly equal when it comes to the accuracy of their responses, but the difference is that Google shows its work. It's fairly easy to look at Google's answer and see whether it came from a reputable source or a letter salad domain. ChatGPT has no such mechanism to confirm the accuracy of what's it's telling you but both its correct and incorrect answers are presented with the same level of confidence.

I would also suggest that ChatGPT won't be able to compete fully with Google until it can handle requests for commercial information. Currently ChatGPT demurs if you ask it which appliance to buy, wheres Google turns up all kinds of results. If ChatGPT does wade into these waters, it will be both under heavy scrutiny from promotive companies while various interests try to figure out how to affect ChatGPT's results.

ChatGPT has so far been an impressive technology demo but I question whether they will be able to really make it into a substantially useful product. I think it will be far more complicated and far more expensive than is generally expected.

Or, as ChatGPT says when I asked:

Overall, making ChatGPT a resource for researching consumer information would require a significant investment in data collection, preprocessing, and model training, as well as ongoing maintenance and improvement to ensure that the model remains up-to-date and relevant.

That raises the question, is it more cost effective to have editors poor over the text, add markup, and run it through QA versus simply hiring someone to read the text? Perhaps a self published author would spend the time on their book to optimize it for audio, but it seems like there is still plenty advantage to hiring an experienced narrator.

Speaking for myself, I wish that DNSSEC was offered as an alternative. I think it could be a simple as placing the self-signed certificate hash in a TXT record. Such a system wouldn't even need to replace the current CA approach, but it would provide a low-cost alternative for those who wanted to self sign.

I will point out that congress's anti-robocall was just overturned by the Supreme Court because it detailed what constituted an autodialer. Since modern companies use a different approach to dialing, it was found to be no longer valid.

I think technology is just hard to legislate because it changes so quickly.

This is a point that frequently gets ignored, in my opinion, when discussing "rock star" programmers. It may be the case that one coder can write code 10x faster than their peers, justifying a higher salary, but it's even more important to consider how much technical debt their code has. Buggy or inefficient code can waste countless hours of developer time for years to come and rewriting bad code that's already in production takes far, far more time than simply doing it right the first time.

Some years ago, I was tasked with fixing some buggy and inefficient code, only to find that its approach was fundamentally a bad one and that it would have to rewritten. The new design was much simpler but it took weeks of data processing to unwind the consequences of the original design. The original coder was long gone from the company and will never know (or likely care) how much difficulty their design caused.

Speaking as another long-term code maintainer, my irritation is that management tends to give new hires the new development work because they don't want to train them up on the existing code base. What happens though, more times than not, is that the new hire throws together something using their own techniques and coding styles, then quits without leaving behind any documentation. At that point, their code becomes my responsibility and usually I end up depreciating or rewriting it in order to bring it in line with the rest of our code base.

I have the strong opinion that in order to be a master coder, one must at some point in their career be in a position where they have maintained their own code for half a decade or more. If a coder never does this and constantly switches jobs, they really have no idea how good their code is. For all they know, their work is discarded or rewritten once they walk out the door.

Does anyone know what the lifespan of the COVID19 virus is in direct sunlight? I read in one article that it was around 30 seconds, but it was mentioned in passing, and I'd be curious to see some scientifically backed literature on the subject.

Specifically, I'm curious how much simply being outdoors affects the spread of the Coronavirus. Perhaps sunlight isn't powerful enough to kill the virus immediately after a sneeze but I would imagine it would quickly disinfect outdoor picnic tables and other surfaces. I'd love to hear from someone with some knowledge of the subject.

For example, here in Colorado, an appliance store that sells furniture is allowed to remain open but a furniture store that sells appliances is forced to close its doors. Rules like these are frustrating because they destroy livelihoods without doing much to reduce risk of infection.

My feeling is that the government needs to transition from defining "essential" businesses and instead mandate a standard set of regulations that all businesses must follow. Dictate number of people per square feet, mandatory protective gear, etc and let the businesses work out for themselves how best to follow them.

There is something strange going on there, but not quite what you think. "Donald Trump dementia" doesn't autocomplete, nor does "Ronald Reagan dementia". I would postulate that Google blocks the term "dementia" for all proper names.

It isn't quite that easy since changing any aspect of the file, even the EXIF data, would alter the file's hash.

A more sophisticated image fingerprinting scheme could work. Google clearly has technology like this that they use for their image search functionality (it's effective even on images that have been cropped or recolored) but I'm not aware of any open solution that can do the same thing. Still, it's a much better idea for solving the problem than just about any other option I've seen mentioned.

These higher frequencies could be perhaps useful for selling stationary connectivity; basically home Internet for anyone who's willing to stick an antenna on the side of their house. Even then though, bad weather could easily disrupt service and force modems into a lower frequency range.

I suppose, in that sense, the sale might be self correcting. That is, the FCC sells the frequencies, the cell companies buy them if only to keep them out of competitors' hands, and no one ever uses them because they're just so ineffective. If that's the situation, the satellite providers just don't have anything to worry about.

The article links to this location where a user an enter their own custom prompt and allow their "AI" to generate additional content.

Enter a few custom prompts and you'll soon get a sense for its strength and weaknesses. It seems to do best when prompted about some current news - presumably it has a wealth of source material to draw on. It's far weaker when the prompt contains fictional proper names.

Though it isn't stated outright, the involvement of Ring makes me think that the goal was to get confirmed video of someone stealing a package via a Ring camera, that could later be used in Ring advertisements. Ring has been forced to take down some of their recent ads because they starred individuals who were not confirmed to be criminals or confirmed to be engaged in criminal acts.

Speaking for myself, I would prefer for there to be a bold line between law enforcement and corporate marketing.