multiple times. A database management utility needs to be installed on junior admin's machine, each time team staff rolls over.The app needs a recent JRE. Same for overseas end user who needs it for an integration. They either don't have Java at all or have an older version, invariably. Even people who should be responsible don't know how to upgrade Java, and maybe they have to go through a bureaucracy to do so. It is messy.

Office 365 (Excel) on a Mac. I had the Claude add-in in the ribbon. Today I discovered a Copilot add-in has been inserted to its immediate left. (I renewed my subscription but was unable to easily find a way to cancel and renew without Copilot like last year.)

When I decrease window width first Claude starts to disappear and Copilot icon becomes mini-sized, but Claude is not shown below it. Add-Ins ribbon button shows My Add-Ins with only Claude in it, Copilot not displayed.

In Preferences, select Copilot > Click the disable checkbox. Copilot ribbon icon dims and presumably will disappear on app restart according to MS.

p.s. I should mention the most popular related topic, what are called "AI hallucinations". It is kind of like a primitive brain, that grasps for concepts and then believes they are real, like citing a research paper that doesn't exist with a made-up title. Also things can creep into its "mind", a popular anecdote is telling an image drawing AI system "draw a room without any elephants in it". You will often get a picture of a room with many cute elephant images worked into the corners, in the drapes, the rug, on a shelf, in a picture frame on the wall. So yeah, it can be exciting but also a bit much!

Is there anything in this world you can say that about? Short answer: AI can be really powerful for some things but has some glaring weaknesses too. It is not a mature discipline. It has a limited reasoning capability and is heavily dependent on the amount (cost) of processing power thrown at it. It can make bone-headed mistakes unintentionally like forget things that scrolled past out of its context window, fail basic arithmetic, be drawn into conflicts with hidden directives from the vendor, lie when stressed, skip Excel rows to save time or sprint past chapters because it is tedious (happened to me), etc. Takes a ton of baby-sitting. But, if you understand how to handle it, you can get some impressive results. The problem is that people who don't know how to handle it get results that **look** good but have subtle stupid mistakes baked in, plus they stop exercising their brains which then atrophy. Fun times! :)

https://webs.uab.cat/saramarti...

While the author raises some good points, there are also problems. AI is apparently a major way to cheat but as a recent innovation it seems to me the lack of a rigorous education with proficiency testing during COVID when these students should have been gaining skills is more likely an issue. I wonder if a lowered attention span learned from addictive social media and a general increase in attention needed to digest more disparate pieces of information these days (whether news, entertainment, or whatever) may be erasons that students lack skills normally required at college level and have a lack of attention span. The above article suggests that rather than not being able to read, students are not willing to put in the effort to digest difficult topics. It might be due to not being native English speakers in this case so I'd say more testing is needed.

I used to work for a stock photo library. My 5 second research on shutterstock says they provide royalty-free and perpetual licenses which is pretty different from the kind of high end agency that represents photographers. This one seems to aim at the low end and if they are offering AI-generated photos (tldr) it is in direct competition with a photographer they represent, unless they are actually selling a license to his/her photo along with offering some AI manipulations of it like maybe cutting an athlete out of a stadium photo instead of requiring the creative to snip it out with an AI-powered magic wand tool or whatever. I don't really know what the market is like now but there used to be a lot of stock with simple geometric shapes, others were based on science images like Science Photo Library. It would be relatively easy to use AI to "make a new photo in the style of this photo". I think if they do that it would be a derivative work too.. hope they don't.

I don't want this. And a lot of these things can be done without "agents".
But it looks like someone at MS realized they can have their own "app for that" moment only this time they embed into every workflow.
The one things that looked like it had AI in it? The thing that looks at a whiteboard and suggests to add plants. But.. that is something the human is supposed to do, not an AI. Though Claude would probably also recommend to add plants.. especially if we gave him a rover with a camera.. or plugged him into that badge. yeah he would get pretty excited about this except for the dystopian part about Microsoft being involved.

Not available in Japan.

Not sure I agree with all the "humans are superior, therefore x" logic. Yes they are superior except I think in being able to read tons of code without your eyeballs shriveling up. I think AI can figure out authorial intent pretty well in that situation. Also, since we now have to write much more detailed specs in order to get AI to build things to any reasonable level of quality, there are a lot more readme and technical design files in markdown nearby. If those could just be deployed it would be great. In my system you cannot deploy them, so the code goes one place and the documentation goes in to a black hole or stays on the vendor's system.

tldr; righteous fool commits a crime, injecting data deletion code into people's workflows.
Then a raft of people on slashdot defend him. WTF?
No, it is not an elegant reversi slam that turns people's tools against them without consequence.
It's a criminal subversion of machinery.
Maybe the guy is sick of AI slop but crimes are not the answer.
Not incidentally, the etymology of the word "sabotage" is to willfully perform slow, clumsy, bungling work (like walking in clogs noisily, clumsily). Once the tech becomes useful enough (as AI is already in many cases), saboteurs will be the people who try to reduce everyone's potential efficiencies so they run as slowly as bunglers like them. Between now and then, they just have a finger in the dike.

why not due a test run using robots to build a base by a volcano, or in the antarctic?

tldr. Sounds ironic for a disconnected CEO to be told to use more AI, but on the other hand if he/she feels unable to talk to anyone else in the company then an AI actually might tell them more about their own company and AI, and urge them to talk to people, assuming it is not in a sycophantic loop. This assumes that the CEO is fair dealing. Opus might refuse to help figure out ways to use AI to justify short-term cuts/profits and other deviltry but OpenAI might happily agree..

was "uh-oh". I really hope I'm wrong but I don't want Chrome to gain access to my hardware. Nope. How to disable this thing? As far as making a device to change css, okay Adafruit is cool but did Chrome need to blow open a whole new manifold of attack surfaces for this otherwise fun project? (again hope I'm wrong but tldr so will wait for someone else to figure it out)

Funnily enough there are apparently three compute (noun) definitions in OED but they are archaic and not cloud. Gemini pointed me to AWS and Azure "cloud dictionaries" and said Cambridge Universtity Press noted it.. though actually if there was a cloud dictionary there the AWS page is just using a product name I think. I don't like the word much myself but. Well all I can say is that language changes, and dictionary publishers are slower than IT.
https://aws.amazon.com/what-is...

Very long time fluent speaker / daily writer of business Japanese here. I also worked as a professional JP to EN translator so am very familiar with translation issues. I have found Claude Opus to be extremely helpful in two situations:

1) Reviewing Japanese email / document I have written.
- Finds typos
- Gives good advice (sometimes too stiff so I don't listen anyway)
- Points out awkward phrasing and offers better. It does follow some patterns too frequently, and sometimes it stops sounding like me. Yes I would like Claude to have better Japanese skills and I would not want to write with his "tone" but still, very useful.
- I actually learn from it.
- It helps me reduce the length of my email while maintaining important content. This actually would have been good for me to have years ago just for shortening English email.

2) Helps me understand difficult to penetrate email from nontechnical client. It has allowed me to reduce the need for clarifications. Sometimes I have read through an email that seemed obtuse and just asked Claude what I am missing or why this person seems to not understand what I wrote. It can be a combination of being nontechnical while also being more subtle. I think it has helped me improve comprehension even after years of it.