This web-based development utility looks interesting and I tried it just now with this prompt;
"I want an Android app that can scan for local WiFi access points when a button is pushed and list them."

It gave me a nice tab completion, "It should also display the signal strength for each access point" which I accepted. And yow, it built the complete app in under 10 minutes. It's running in an emulator in the browser and showing mock access points, visually looks very good. I am officially giving everyone permission to use my amazingly awesome prompt, try it for yourself.

When I ask how to install the app on my phone I get; "To run this application as a fully functional, bare-metal hardware app on your physical Android phone, you will need to build the provided native Kotlin source code using Android Studio" and there are some pretty complicated instructions. That will probably be a roadblock for most people and it would be a PITA for me too.

However I'm seeing that Gemini is now integrated directly into Studio which is extremely handy. I can probably use the same prompt there and get a complete installable app in one step.
https://developer.android.com/...

"was arrested thanks in part to the cameras he allegedly sought to destroy", always approach your objective from the rear before applying your cosmetic enhancement.

Sounds like OpenAI is really struggling. They've already suffered a lot of brain drain, and in fact Anthropic was founded by former employees. I'm suspecting low morale, distrust of Altman, increasingly bad reputation of the company, etc. If I were Karpathy I'd want to play on a winning team with a better attitude too.

I find that everything is easy and ordinary if you know how to do it, and there is no shortage of nerds in this world. My point is that the tech is readily available to those who are 'practiced in the art' and it isn't a high bar, there's no stopping it.

Meanwhile, today I noticed a new Flock camera being installed on a road I travel every day. It's a choke point at one of the entrances to town, the police will have a timestamped record of every car entering and leaving. I'm tempted to accidentally hit it with some spray paint.

You are correct about the resolution, my bad. But in my experience a camera like that is able to capture legible license plates of all passing cars when it's mounted next to the road, like the Flock cameras are.

Don't believe me? Not a problem here. Maybe its best that people will think license plate reader tech can't be assembled by a hobbyist using a few parts you can get on Amazon.

So you really aren't a part of the cohort under discussion, right?

The teaching profession may presently be less vulnerable to AI infiltration than others in terms of required qualifications and how you operate. Most other white collar jobs are going to want to see you produce work outputs with AI assistance and be good at it.

5k resolution isn't anything special these days. I own a few of these cameras, dirt cheap and very capable both day and night.

https://www.amazon.com/dp/B08L...

>> I'm doing just fine

Are you a recent college graduate who found a job?

You can get a good wired 5k IP security camera on Amazon for under $60, and receive the video stream on your laptop. Higher resolution wireless cameras are not much more expensive. Mount the camera near the road and point it parallel to the direction of the vehicles, similar to the way Flock cameras are positioned, and it will get an adequate image of the license plates. No problem.

The problem is that what Flock is doing is not very hard to accomplish these days. Anyone can mount an ordinary IP security camera in a place that has a good view of a street, and pass all or some of the image stream along to some pretty inexpensive AI for analysis. AI that's capable of picking out license plate numbers and all other identifying features of the vehicle.

Flock has a nice solar+battery+cellular internet package for this that they mount just above easy vandalism height. It is well designed and they get a pretty penny for it, but it basically just does what I described. Not out of reach even for a hobbyist.

Strong words there, but I agree. This tech is something that people are going to have to embrace in order to succeed. I'm way behind the curve on AI expertise I'm sure, not orchestrating a team of agents etc, but I use it every day. I've learned how to work with AI assistance effectively, and that's becoming an essential skill. It can turbocharge everything you do.

People in college should be getting a heaping helping of AI-enabled project building. Practical use cases. Come out of school with a set of smart AI agents working with you? Immensely valuable.

There was a time when OpenAI appeared to be ascendant. So far ahead of everyone that they were almost a sure bet. That's much less obvious at this point.

Apple realizes that it would be foolish to hook themselves up exclusively to one vendor. There are loads of models and agent frameworks out there. Many of them are very good, and well up into the competitive rankings. These days your prompts pre-screened by an agent that sends each one to the most appropriate other agent that's most economical and capable of handling that specific situation. It might very well not be an OpenAI solution.

Apparently CERN doesn't actually use KiCAD. "The libraries are the result of automatically converting the original Altium Designer source libraries". And "3D models and datasheets are not included" which is unfortunate, it is very handy to be able to see a 3D rendering of your layout.

>> How many tons of CO2 have you personally generated

How many do you think it is, smartass? The number of tons I've used is apparently zero. A full day of AI assistance prompting uses about as much extra energy as running the dishwasher. Typical prompts generate to 2–4 g of CO2, I get a lot of work done with 100-200 per day.

https://www.cnaught.com/blog/h...

"A standard Google search uses approximately 0.3 watt-hours of energy and produces roughly 0.2 grams of CO2. By most estimates, a ChatGPT-style query uses somewhere between 3x and 10x more energy than a traditional web search. Generating an AI image consumes roughly the same energy as fully charging a smartphone. To put individual impact in perspective, analyst Andy Masley calculates that it would take roughly 1,000 ChatGPT queries in a single day to increase a typical American's daily emissions by 1%."

If you look at the jobs website you'll see that they are all senior positions requiring lots of specific experience. Companies like GM do little to no IT training, so if you became complacent your skills are obsolete.