RIP Rickey Henderson. The greatest lead off player of all-time.

... On Windows, press 'Alt + Space' to open the chat bar, and on macOS, press 'Option + Space'.This allows you to interact with ChatGPT without opening a browser window."
Not great. Yes it did ask if I want that answered by chatGPT. Siri should use chatGPT to enhance not distribute work.

I just installed macOS 15.2, I gave siri perm to use chatGPT, I then click on siri to start and typed, "Is there a quick key I can use to start you?", ans "Yes, you can quickly access ChatGPT using keyboard shortcuts....

But I think the key here is that the madness that is this murder has an origin and is understandable, but of course it is not justifiable. Personally, I think this is a terrible way for society to solve problems, but we should be listening to people and how they react and feel to events like this.

So when we see a horribly thing happen to a person that is part of the system that makes lots of people feel helpless and for lack of a better word screwed, they try to justify it.

How many times have we heard or read "I don't condone murder but that CEO was a horrible person" or something like that. My take away is that the temperature or pressure that people feel from the healthcare industry is rising and it is a terrible part of life where people feel helpless.

Which, then one could say that we are the AI in the AI. Anyways, If we continue to call LLMs AI, my guess is that we will start to resent the term AI for garbage and business speak. It's promise is too broad.

But, if it doesn't find the context, or is something that doesn't have many examples in the train set it will fail. AI would be able to search the outside world for data to coerce into a training set and then build a separate model it to use.

If there are enough examples of something in the models train set it can very accurately generate examples and the generated examples don't have to be exactly like what is in the train set.

Don't get me wrong, it is impressive and astonishing that it works. Take language, chop it up into tokens, set up a large complex network, do some very expensive training and you get a very good language generator.

It is possible that in the future, possibly the near future, LLMs will be an important component of AI, but as it stands it is an auto - regressive model, setup for prediction.

I wish we would stop calling the body of technolog that we call AI, AI. LLMs not AI.