I'm thinking very very stupid. No sprues for casting.

Agreed. This is all stuff that at MOST should be accessible over the LAN. The ESP32 is cheap and provides the WiFi and enough power to run a simple RESTful web app. If I actually need/want to access it remotely, it'll be through a well protected integrated web servie on a jump box.

A cheaper manufacturer could probably make the ESP32 do double duty as the primary micro-controller with a suitable interrupt routine.

The situation was worse than I thought. According to the AAIB report, they tested the material and found the glass transition temperature to be about 53C, so the jackass printed it in PLA.

Also, since it's Poly Lactic Acid, our metabolism is already well equipped to deal with the breakdown product.

It's great for things that will be indoors at room temperature. I made a snap-on camera mount for my monitor that has held up great. But if it was a dash mount, it would need to be PETG or CF-PA6 to not sag in the summer.

Getting their jobs sent to American AI won't be noticeably better for American workers than getting their jobs sent to Chinese AIs.

According to an AAIB Field Investigation report (pg. 4), two samples from the intake were tested and found to have a glass transition temperature of 54.0C and 52.8C

So some idiot printed them in PLA. PLA is great but is very much NOT temperature resistant. It has been known to sag in a hot car.

Now look at the ratio of human driven cars vs. Waymo cars.

The problem was using a cheap substitute part. I'm guessing an injection molded ABS part would also have failed in that scenario.

CF-ABS is NOT like fiberglass at all. The CF is chopped into fine bits. They lend some stiffness at room temperature but not strength to the part. Certainly the carbon fiber bits don't lend any heat resistance.

That's why I suggest a mitigation to the increases for industry based on local employment. Data centers employ very few people per-Killowatt and so contribute a lot less to the local economy compared to those other industries.

It would make sense in conjunction with an employment based mitigation. Data centers employ very few people once operational (they're not called lights-out facilities for nothing), so no mitigation. Major manufacturer provides many steady jobs, more mitigation for them.

Of course, things get complicated. There are mini data centers being set up in people's back yards where the waste heat warms the home owners house. That doesn't employ a lot of people but gets effectively double use of the energy for at least a good part of the year, offsetting other energy use, so it should see some form of mitigation as well.

The bigger question though is how long until the data centers are abandoned? The big AI companies and their investors are operating at a loss as they jocky for market share and train ever larger models. But will people actually find the AI useful enough to pay for it once the investors start demanding their ROI? Will managers come to realize that they might be better off hiring people suffering schizophrenia with frequent psychotic episodes?

End to end encryption, for a toilet? Frankny I do not want a TOILET to connect me "end-to-end" with anybody. They're doing it wrong.

Time to pull up the sheet on IOT. Not only has it gone up it's own backside, now it's trying to go up ours too.

How about "hit", the most common word used to express the concept.

I'm not so sure there have even been enough incidents to decide if Waymos are more or less likely to run over an animal, but I *STRENUOUSLY* object to trying to soften it with 'made contact with' in the press release.

What seems most depressing about this isn't the fact that the bot is stupid; but that something about 'AI' seems to have caused people who should have known better to just ignore precautions that are old, simple, and relatively obvious.

It remains unclear whether you can solve the bots being stupid problem even in principle; but it's not like computing has never dealt with actors that either need to be saved from themselves or are likely malicious before; and between running more than a few web servers, building a browser, and slapping together an OS it's not like Google doesn't have people who know that stuff on payroll who know about that sort of thing.

In this case, the bot being a moron would have been a non-issue if it had simply been confined to running shell commands inside the project directory(which is presumably under version control, so worst case you just roll back); not above it where it can hose the entire drive.

There just seems to be something cursed about 'AI' products, not sure if it's the rush to market or if mediocre people are most fascinated with the tool, that invites really sloppy, heedless, lazy, failure to care about useful, mature, relatively simple mitigations for the well known(if not particularly well understood) faults of the 'AI' behavior itself.