I also make nerdy jokes/puns, and I had similar experiences with beneficial model steering that such puns create.
It's interesting how more loose/informal prompting achieves good results, there is some cultural understanding in the models from training. Once I asked the clanker to remove the gambiarras and puxadinhos that it wrote as part of an experiment, and it promptly fixed those.
I have zero quality issues on Z.ai Coding Plan and maki.sh as the coding agent, but I've seen many reports on 3rd party providers that they host heavily quantized versions of GLM models.
Happy to disagree with you. There are many examples of the US using the state espionage machinery for private purposes. The extent of "private" is debatable given the US as a nation behaves more like a big corp.
For me it depends, if I simply take the answer from the LLM it goes straight into the shortest memory neurons my brain has.
Now, if I go back and forth with the LLM to say, taking the language learning example, to explore the etymology of the word (which for me is far more interesting than the translation itself), then I learn a ton more.
Exactly! It got so easy nowadays to use AI to setup scenarios since the busy work of writing code/test harnesses and setting things up for the benchmark is done by the machine. Then throw away what does not work.
Some benchmark that would take weeks to plan, code and set up is now hours and days - the time is now spent on the benchmark itself, not on temporary code.
For me, them allowing API usage on coding plans so we can use any harness, and returning the full unabridged reasoning back are how they earned my trust.
Weird. I have seen it asking the harness to do `find ~ -type f | grep` to try and find my agent configuration .json file when I asked it to add a MCP server. Stupid, but they weren't sending the files back home. This was with older models though. Newer ones are a bit smarter than that.
I assume they can spy on anything, that's why I refuse to do any kind of BYOD that requires MDM enrollment. It's also one IT mistake away from wiping out my personal data from my own property.
Company's Outlook, Teams, MS Authenticator and Slack on my personal phone? Ehhhh fine.
The second IT requires a profile it's the second I'm uninstalling anything company related from my phone. They can give me a company phone, or I will only access work related things from their laptop.
Happened twice to me, one time I got a company phone, the other, computer only.
And then some colleagues even paid out-of-pocket to upgrade their Android phones once IT dictated their OS were out of support. No way.
My first phone that had a (rear) camera 20 years ago also had a chrome-plated mirror thing to help us take selfies. I guess nobody called them selfies then.
I guess point and shoot cameras also had those mirrors back then.
Glad to know that kids rediscovered camera selfies.
Nah, I live in a major city between two airports, under an airplane departure and a heli route. I can't really launch rockets or balloons from my home.
When I was at uni I didn't have the funds for this hobby unfortunately.
It's interesting how more loose/informal prompting achieves good results, there is some cultural understanding in the models from training. Once I asked the clanker to remove the gambiarras and puxadinhos that it wrote as part of an experiment, and it promptly fixed those.