Good progress is being made on Xfce’s Wayland compositor, which is really exciting to see. Might spin up a VM later to try it out!
Hello Andreas,
Just want to mention that I would be very interested in hearing how well that tryout went.
It may be a deciding factor for me to move away from MATE to Xfce, however loathe I may be to do so, having tried Xfce previously.
Of course! I hope I find some time later this week to set it up.
Another good read while there:
He brings up good points on LLM training and whether code output by LLM’s is original or copy pasta. And to be fair, it’s getting to a point where.. Can you even tell anymore?
Personally I find coding agents amazing and look at them as another tool in my box to help me get things done.
Agreed @Plyply Like with most things, you get out the effort you put in.
A lawyer has been showing me how he’s been using AI to completely automate certain tasks and customer communication as well as writing approaches or whatever they’re called. He said: “you know Hayden, what this technology will do is expose the bad law firms/lawyers and elevate the good ones”.
The next week he shared a video online of a judge exposing a lawyer who used AI for his research and cited several cases that don’t exist. lol
This is because if you know law and know your niche, then the advantage is that you, the human element is the only qualified entity to identify errors. But if you are not well qualified in your niche, or using it like a search engine to ask questions on things you don’t already understand, then you tend to depend more heavily and just accept what’s provided. Besides its a tool for serious use, not a search engine.
So agreed, I can only imagine how scientists, engineers and experts in various fields can now achieve things FAR beyond what they would have had the time, resources and funds to invest in without AI. But due to their professionalism are able to ensure the heavy oversight and redirection of time and energy goes into going over things instead.
It’s like being able to jump and reach things off a shelf that’s 8 feet above you, now with AI you can jump and reach things 80 feet in the same amount of time BUT, the catch is, that once grabbed you must take the long way down to ensure the safe and accurate arrival of what you have now achieved.
The problem is far too many people and companies have bought into the idea and fear that AI is here to replace us in the workplace and beyond. So much so that companies have been even trying to do just that and are learning now the hard way! lol:
But as you said it’s a tool, a tool for us to use and get the most of if we are willing to still put in tons of effort and human ideas and innovation. Because its far more powerful IN our hands, than left on its own and like other technology we will always be needed to amplify it.
Originally I wrote this at the end of 2023 and it think it’s even more true today:
I agree—to add an anecdote from my field of work, neuroscience, machine learning tools help us to make sense of large data sets and extract information much easier. For example, we use models to extract pupil diameter information from videos. Before, each frame would have had to be annotated, or a less reliable metric would be used; however, thanks to convolutional autoencoders, we only need to provide a minimal dataset and the model can extract pupil information from tens of thousands of frames in tens of minutes.