GarryP

I do indeed think they need to shift towards real skills, the kind of skills that survive automation and are in short supply. Hairdressers are the kind of thing you can train up locals to do. And you are going to want to keep any such roles to absorb the tranche of unemployed.

As I pointed up, I don't think retraining is going to work generally, because the kind of roles you can train people up to do quickly are the same ones that are likely to be automated.

And the speed of implementation will be fast, within 2-3 years at max. That's because wages and dealing with people takes up such a slice of costs. An automated supply is going to be able to undercut the cost price of a manual supply - so those that don't move fast will be put out of business anyway.

BadgeIsBack

Are they going to automate all those people who believe the sky is falling? :-)

I think there will be an adjustment. I still think you have to employ to people to fix all the mistakes and manage the errors and fallout from automation.Working in IT I know that there's only so much you can do. In the real world, it is the edge conditions you need people for. All this software people have built over the last 20 years and admin people still need to sort..only today a client had to spend time reconciling the efforts of machines...surely we have learnt that machines don't always get it right..project mgmt is still an arcane discipline..

GarryP

So you employee a tenth of the number of people to deal with the edge conditions - or you simply don't bother with them at all.

If you can deal with 99% the general car insurance, but can't automate the 1% edge cases - you just don't both with them.

Beoz

I work in automation and all its done is increased the amount of people required to create, market, maintain, develop, customise and increase the automation.

It was never about removing people. It was all about increasing productivity.

BadgeIsBack

Good man: it reminds me of good old software dev. We have gone from cobol to immensely verbose OO languages, framework libraries, toolkits, QA and configuration. It has got more and more complicated infact.

Beoz

Forget the development part. You can have the best code or best software in the world, but without it being used, used to its full potential, and repeating that use year on year, its worthless.

Shard

Fairly good AI YouTube with a panel discussion including one of the Googlers that is working on autonomous vehicles. He noted that the AV is so far very environment specific (basically city rather than country) but obviously they are refining that. His timing estimate for roll out of AV was "within a decade". So it sounds like 2025-28 for roll out? Can find the link if you are interest, or if you look up Neil Tyson Asimov lecture (very recent) it should come up. He commented at about 32 mins.

ozzieeagle

Elon Musk's plan for 11,000 satellites to cover the world with the Net has me wondering.

Skylink eh ?

Bit uneasy about this one. No hiding place is the first thing I think, rather than full planetary coverage.

GarryP

Within a decade is also in two or three years time....

With Ford and GM both saying 2019 now, they are either wildly optimistic, or in the know. My guess is that they will try and do it slowly, so as to not upset the natives, but would be aiming to have significant coverage 3-4 years after the initial roll out - and using the example of success to knock down silly local laws. From there I would expect it to kill second cars, and then insurance would kill primaries.

So rollout/replacement of autonomous vehicles within a decade.

GarryP

There are quite a few services in the wings, not just Elons.

And if you are really worried about "no hiding place" you should be worried about governments 360 degree coverage.

As it is these services are likely to make a mess of the NBNs satellite and fixed wireless services - lower latency, greater availability, etc. They will work better the less dense the population, so the country will likely get a better service than the city (an interesting change).

Shard

It sounds like it will be a gradual introduction of autonomous features (eg. braking) over a period of years until the public gets more and more familiar and comfortable with the concept. The Google guy (Waymo, actually) said he didn't think there would be any big bang in AV.

GarryP

Nah, I doubt it.

The point is the money is in full autonomy, cat 4 at least. cat 1 & 2 are already here (eg braking) but it's not till you can get rid of the driver that the money starts rolling in.

The way I think they will ease it in is via 'pilots' in certain areas, that then expand out to encompass the entire city and quietly drops the 'pilot' tag. Before you recognise it, 30% of journeys are autonomous and Springfield is doing big bang launches to catch up to Shelbyville.

That will be for robotaxis. My guess is that personal cars will start with autonomous operations on motorways/freeways, and then work into the commute.

Shard

Pilot intro model sounds feasible, especially ex-UK/Europe where the street are on a grid and have wider spaces. It would still need the Cat-4/5 AV which is what seems to be a number of years off.

GarryP

https://azbigmedia.com/autonomous-ve...vation-valley/

Shard

Interesting. These are the pilots then.

It also noted: Gaffar says fully autonomous vehicles
for the masses won’t be happening any time soon but that doesn’t mean we can’t make the current cars on the road smarter in the meantime.

But I guess it could always change on a dime if the tech proves itself and the public demands it.