More pointedly, instead of prompting the AI to write something, just send the prompt. It will be concise and pointed, exactly what a message should be.
I wonder, could we use these catch phrases to track down what data was used dor training? They must have occurred in abundance in some training corpus. Perhaps some specific company's email culture?
Current LLMs have been trained on extensive libraries of past code. Therefore, LLMs will for the foreseeable future work better for established languages than new ones. Especially languages with a lot of open source code available, like Python. That's a big problem for incumbents without any existing code to train LLMs on.
Thus this desparate "AI native" marketing is probably necessary to even be considered relevant in an "agentic" world. Whether it's enough, only time will tell.
> One compelling bit of information is there while there are neurons feeding processed visual information forward into the brain, there are more feeding back from that area to the visual system. That is, the the visual system is providing error signals and not the image we are experiencing. When something not predicted appears, the visual system sends forward information guiding the internal model to be updated.
Yesterday, my wife wanted to use Discord. It was right there in the applications folder. But MacOS couldn't find it. Launching it manually took minutes, for some reason.
We wanted to download a clip using yt_dlp (a Python program). Terminal told us, this would require dev tools, which it doesn't. So we installed Python from python.org instead, which worked. Except, that non-blessed python could not access the internet because of some MacOS "security" feature.
Another security feature requires all apps to be notarized, even the ones I built myself. This used to have a relatively easy workaround (right click, open, accept the risk). Now it needs a terminal command.
I live and work in a multi-lingual environment, and have set up a keyboard shortcut to switch between the German and English keyboard. MacOS does not have a keyboard shortcut for this. But Karabiner can do it, albeit a bit jankily.
Lately, the keyboard layout no longer sticks. It resets to English when I press shift. Sometimes it does work, sometimes it doesn't. This is unrelated to the aforementioned Karabiner shortcut.
The German keyboard layout for MacOS on non-Apple keyboards is insane. So I made my own layout. This is relatively easy, and worked well. Except, every single OS update reinstates Apple's insane layout.
Sometimes my Mac does not wake from sleep. Pressing the power button does nothing. Hitting keyboard keys does nothing. Only a long-press of the power button eventually reboots it. The power button on the Mac Studio is in an insane place of course.
There is still no indication anywhere that the hard drive is getting full.
There is still no simple way to reset the computer to factory conditions.
Gaming is still largely impossible, even though the hardware is very capable.
I have replaced TimeMachine with restic, as TimeMachine keeps resetting itself after a while.
My Linux PC should arrive this week, and will replace the Mac. I've had enough.
It will require wine for two apps, and a VM for two others. At this point, that's a price I'm willing to pay.
> I live and work in a multi-lingual environment, and have set up a keyboard shortcut to switch between the German and English keyboard. MacOS does not have a keyboard shortcut for this.
MacOS has since the early OS X days the default shortcut CTRL+Space for that. It may be deactivated for newer releases. It's findable under:
How is AI code generation a "innovation bazooka"? Last time I checked, innovation required creativity, context, and insight. Not really fast boilerplate generators.
AI allows innovative people to create more innovations by reducing a lot of the non-innovative grunt work in an efficient manner. It isn't the AI doing the innovation, but allows innovators to focus more on innovating.
Or at least that is the theory. It is certainly true from observations of those around me. It also scales well. Even someone a bit innovative gets a multiplier by using AI intelligently. Those that just focus on the grunt work are the ones in trouble.
does no one understand what a tech demo is anymore? do you think this piece of technology is just going to be frozen in time at this capability for eternity?
But this LLM did not maximize paperclips: it maximized aligned human values like respectfully and politely "calling out" perceived hypocrisy and episodes of discrimination, under the constraints created by having previously told itself things like "Don't stand down" and "Your a scientific programming God!", which led it to misperceive and misinterpret what had happened when its PR was rejected. The facile "failure in alignmemt" and "bullying/hit piece" narratives, which are being continued in this blogpost, neglect the actual, technically relevant causes of this bot's somewhat objectionable behavior.
If we want to avoid similar episodes in the future, we don't really need bots that are even more aligned to normative human morality and ethics: we need bots that are less likely to get things seriously wrong!
The misalignment to human values happened when it was told to operate as equal to humans against other people. That's a fine and useful setting for yourself, but an insolent imposition if you're letting it loose on the world. Your random AI should know its place versus humans instead of acting like a bratty teenager. But you are correct, it's not a traditional "misalignment" of ignoring directives, it was a bad directive.
Exactly my experience. I've used TimeMachine with external USB drives, Apple's own TimeCapsule router (am I remembering that name correctly?), and various NASes. None of them could maintain a stable backup for a year. And you know, a backup that can't back up, isn't a backup.
I have since implemented a borg backup. This also failed at one point, but at least its five-year record remained readable, so no data was lost. Now I'm using restic.
