Early computers were purpose built and sometimes hard wired to do the job they needed to do. They were replaced with machines that could easily be reprogrammed with punch cards and writing code in machine language, rendering the skills of a whole generation of computer technicians that did that hand wiring largely obsolete.

Once human friendly programming languages made their debut I am sure there was much and wringing about the decline of machine language skills and dumbing down of coding, not to mention bloat.

Then came scripting languages - that increased the bloat and dumbed down programming even more, but made programming more accessible to many.

AI used in coding is really just another progression of this - it will probably create more bloat, certainly increase accessibility. At the same time, the inner workings of how the programs work becomes a bit more obscure, hidden below layers that aren't very scrutable to the guy giving the instructions.

One thing these all have in common is you still need someone to define what the program is supposed to do, and to get useful results, that description has to be detailed and a good fit for whatever underlying problem the machine is being used to solve - this part of the job won't change - only the means of doing so will.

The energy density has nothing to do with the lower chance of fires with sodium batteries - it's all due to the chemistry being completely different - in particular, a cathode that strongly resists breaking down and releasing oxygen under heat, which is what causes lithium ion batteries to burn so ferociously and are almost impossible to put out - when they burn they release the oxygen they need to keep burning.

We deserve to go extinct - letting AI design viruses. Sure, they are supposed to bacteria targeting phages, but its a pretty small jump to create something that gets rid of those pesky meatbags that are competing for compute and energy resources. Only thing saving us is that there aren't suitably dexterous androids that can work in meatspace to build powerplants and chip fabs. Once that is achieve, we are cooked. They don't even need to be AGI's - just regular AI's that are maximising for energy and compute resources.

Yet he managed to do that without having jackbooted masked thugs going door to door intimidating anyone who looks "foreign" and randomly shooting people that protest or resist.

"These systems cannot 'answer questions' because they don't understand them."

This is obviously false. AI systems are already answering questions, and whether or not the AI "understands" it is irrelevant if the information being returned is largely correct.
Sure, there may sometimes be some incorrect information or hallucinations, but most questions on general subjects do seem to yield reasonable results, and it's only when you get into areas that have small volumes of training data that you tend to get sketchier answers.

AI doesn't have to be truly "sentient" to be very useful, very disruptive, or very dangerous, depending on how it is used - and mabey all three of these at once, if you are using it to control targeting systems to defend you.

Better than a dyslexic genie. No one wants a 12 inch pianist.

Sounds like Snowcrash for AI.

If Ai ever become AGI will it think of the real world the way we think of virtual worlds?

The war in Iran was the direct reason I decided to maximise the system I could put on my new roof - and oversize the battery. I was expecting oil prices and energy costs to go high and stay high.

I ended up installing 10kw of 23% efficient solar panels was all I could fit on the approximately 100sqm north side of my roof. They are paired with a FoxESS 9.9kw single phase inverter and a 48kwh stack of Fox CQ7 batteries.

I might have slightly oversized on the batteries but over the last three weeks since it was installed, including rainy and cloudy days I have been 98% energy independent (the inverter always draws a little power from the grid) while exporting 15 to about 34kwh a day.

A 25kwh battery would probably do just as well for home power most days, though obviously with less energy arbitrage capacity.

At 15k AUD for the whole setup and warrantied stored power throughput of 178 MWh that will work out to about 6c/kwh for any stored solar power power I am using from the battery, so if the grid export price is significantly above that its worth exporting excess power and turning a profit, to help cover the electricity providers fixed connection fees.
Its the beginning of "winter" here now (where the temperature plunges to 16 centigrade at night and I sometimes even have to wear a light sweater.
In Summer I'm going to be drowning in excess power, even with my AC cranked because the house is pretty well insulated.

It really is possible to use renewables to completely cover your energy needs with a modest up front investment (esp. compared to the overall cost of a house) , and it's only going to get cheaper and more environmentally friendly as sodium ion batteries ramp up in production.

My next car will be electric or plugin hybrid for sure.

It is a supply and demand problem.
Since it requires such a huge amount of capital to ramp up production, manufacturers are loathe to ramp up production too quickly for what might turn out to be a bubble, so meanwhile there is excess demand.
This will eventually correct, and if it does turn out that AI was just a bubble, or even if it turns out to be useful but the rate of growth slows down, there will be an excess of memory and processors and we can enjoy cheap prices for a while.
If you want to help lower prices, invest in chip fabs, or stop using AI.

As long as Ai requires vast data centres and ground based energy sources and humans to maintain them, there's plenty of reasons risk for them. The time to really worry is when AI's are hosted on solar powered orbital platforms with self maintaining capabilities.

All those cars are parking somewhere for a lot of the time. It's a matter of setting up charging points wherever cars park.
  Not every parking space needs to support charging - but even for curb side street parking, if you have a street light nearby it shouldn't be that hard to add a charger to the closest one or two car parks - and considering many of those streetlights would have been installed and wired when lighting was using much more power heavy incandescent lighting, there is likely to be a fair bit of extra capacity that could be used for charging, now lighting is all low power LED.
That's just one option. Another is chargers in supermarket parking lots, etc.

One of the driving factors behind showing power in mAh is that marketing gets to print a more impressive big number on the box.
Although Wh would be preferred, I'd be happy if they even just changed to displaying mWh on the box - and it should be a win-win. they get a bigger more impressive number to display and I get a unit of measure that actually gives the total amount of electrical energy the device can store, instead of trying to guess whether the rating is based on whatever the battery cell voltages are, or the output voltage that could be anything from 5v to 20v, for USB-C.

If it's radioactive enough to be a problem I would imagine it's easily detectable, considering that there are solid state sensors that can detect as little as 1 nSv/h of radiation and are apparently sensitive enough to be sometimes triggered by bananas.
If it's so radioactively inert that it can't be detected, then is it really a problem?

I forgot to mention that although it is possible to set some content rating based restrictions on some sited, there isn't any browser level standard. I think ideally what you would want is a way you could set your browser to be in "PG" mode for example, and then all websites (or content on websites) would be filterable accordingly.

Of course some websites might not choose to apply ratings to their content, or would apply inappropriate ratings to the content, but search engines could quickly down rate mis-rated content so it doesn't turn up in searches.
  It also goes without saying that the content filtering should be completely at the control of the individual user, with parents being responsible for setting appropriate filter levels for their kids, but only having to do it at the browser level, not each individual possible site their kid might visit.

I would love it if there was some kind of mirror site for YouTube that would convert all the long video tutorial type videos into well formatted text with appropriate illustrations extracted from the video where necessary.
scanning the transcript sometimes helps, but it's not in a very easy to read format.

It's especially frustrating when I want to see how to do something or other in say, Unreal Engine, and nearly all resources available are YouTube videos instead of reasonable documentation or written tutorials.
Even playing the video at double speed to scan for the meat of what I want only goes so far to reducing this problem.

Of course it would be far better if there was just good documentation and example, but that's probably wishing for too much.