In the world of social media, highly available encrypted comms, and so on, it's hardly a surprise that this kind of thing now happens. I wonder if we should be a lot less surprised in reality.

Asking MANY people to keep a secret is incredibly hard now.

Maybe we need to rethink the process.

It feels like "Google is an amazing search engine, but how does it make money?" again.

I suspect the answer will be similar, and that's advertising.

There's a LOT more information because of the way the interaction happens.

Data and privacy issues though, and lots of them... but US companies.

This situation is very much like the early 2000s where there were LOADS of tech companies that simply went on hype, and then died.

It wasn't that there wasn't anything there, it was that too many investors threw money at poor investments.

We're in exactly the same period now.

Ask a copywriter if they have a job any more.

LLMs and AI aren't going anywhere. There is a bubble and it will burst. But just like AWS emerged from the dot com bubble, there will be something that arises from the LLM bubble that will likely be useful and valuable and widespread.

Being anti-AI is a genuinely strange stance at this point.

8. LLMs are incredibly useful at text based analysis and many other things. There is huge value in using LLMs for *certain* tasks.

Knowing which tasks is the key.

9. They will be used for things they shouldn't. Twas ever thus with every new technology. Read some history.

7. ENERGY USE IS HORRIFIC. Yes. Very well aware and probably more aware than you are. I've been campaigning around energy use of data centres and the impacts on climate for years. It's not a new topic to me. But energy use will change over time and we need far better data to make good decisions.

6. Not learning about Ai and LLMs is probably a bad idea. They aren't going to disappear and with the huge investments, the US Govt isn't going to do anything because "stock markets go up" is the dumb metric used to determine success in the US.

Take the time to understand when and how to use LLMs.

5. AI companies are basically following the tried and tested "create a company, float on a stock exchange, become the biggest, find a business model". They don't need a business model if people will still invest, and they are. When they stop the bubble will burst but LLMs will remain in some form.

4. Copyright is a massive issue with LLMs. HUUUUUGE issue. But the companies who make the LLMs are mostly US based and until the US government (and states) actually regulates and upholds copyright in any meaningful copyright, then there very little you can do... except vote in better politicians.

3. Bias is definitely an issue, and grok specifically is a real problem here, but I challenge anyone to build any LLM that doesn't have some bias. If you've ever studied the subject you'll know it's impossible although you should absolutely mitigate *harm* where it is found and highlight bias.

Time. Which is also cost. A human will take a lot longer to edit. Also I can try new approaches more quickly to see if that makes sense.

Much as I don't like a lot of the issues of AI (e.g. LLM energy use), there are *significant* benefits.

YAY!

The problem is the lack of consistency across GPs and the difficulty of switching (yes it's meant to be easy, but it's a huge step for many).

We left the "family doctor" idea behind a long time ago, and GPs are in general amazing with what they're asked to actually do.

Well, they aren't, so there's that.

Not saying they're perfect, but that's a lot of over reaction.

It's amazing how something that you think is clear can be misunderstood so easily.

LLMs aren't magic, but they are very useful for specific scenarios.

And always check the output is valid before using it.

Cannot stand how people make websites now. It feels like nobody who has any control understands the simplicity of the web and the minimal requirements of a web page.

Oh absolutely. The problem is that most people don't understand them to know why they are needed (in some circumstances), and wouldn't know how to replace them.

In other news: good to hear from you and hope you're doing ok.