abelard
Player Valuation: £35m
Do you have a link? Cheers
Current Affairs Robotics and AI....
- Thread starter peteblue
- Start date
- Status
Do you have a link? Cheers
Black Belt Jones
Player Valuation: £25m
Call centres seem ripe for this sort of disruption, and I believe they are quite large employers of people with minimal qualifications. The software is there now, albeit it's rubbish, but it doesn't seem a stretch to see that getting better v rapidly to replace these workers.
I agree things like automated HGVs up and down the M6 seem a bit of a way off, even if the technology is nearly there right now, it's still a major societal development.
What we do if this does all come to pass in terms of supporting people is an interesting question. Assuming we're not going to start shipping skip-loads of folk off to the biofuel plant, and capitalism needs its consumers, what's it going to be? Monthly allowance to play video games?
Do you mind providing an example of a task that ANNs have yet to do at an exceptional level?
The vast majority. I've spoken to hundreds of chief data officers over the past year, and practically all are doing the square root of naff all with AI. The odd prototype and test here and there does not a revolution make. Ask yourself this. It's been 18 months since AlphaGo beat Lee Sudol and a whopping 8 years since Watson won Jeopardy, yet neither IBM or DeepMind have generated much actual revenue from the deployment of these seemingly cutting edge technologies. How long will it be before Waymo release a vehicle to the market?
Thing is most people confuse Automation and AI. You are certainly correct people are not losing jobs to AI not as much as people seem to think.
Automation the more technology progresses the more it is used.
Automation is still taking jobs.
I have been part of six major projects in the last five years. We pretty much have helped facilities (various companies) build fully automated systems that then means they have no use for manual labour.
I think in total the companies i have consulted for have put at least four or five thousand people out of work.
AI is a long way off but again people cannot tell the difference between it and automated systems.
Just like people calling military drones robots. Technically they are not. They are unmanned but are definitely 100% controlled by a Human. Robotics and its use in Automation seems to always confuse people too.
I would place the blame concerning a lack of autonomous vehicles on the legislators that seem reluctant to embrace the inevitable changes we will face over the coming twenty years.
Before my breakdown, I was doing consultancy on behalf of a sizable AI company of whom I contributed to during my undergraduate thesis. At the time, we were in talks with one of the largest manufacturers in China about automating a sizable chunk of the production line. The only limiting factor then was the sheer upfront cost of purchasing the robotics that would have enabled the project to develop, but the technology is there.
The only real technical issues I see right now is the comprehensive lack of inter-operative data in which can be effectively used to do massive wide-spread analyses. As you are aware, I mainly spend my time these days working with the NHS to get their clinical informatics infrastructure up to scratch as to enable us to really explore the nuances of public health issues. One of the things that has excited me over the past year is an acceptance that ICD-10 and SNOMED CT aren't anywhere close enough in terms of truly exploiting the mass levels of data the NHS is able to produce, and that we take the right steps from now on to ensure that we're better placed to do so in the future.
The same thing goes for industry, and I feel that the market forces are there to ensure that it does happen.
Of course, in the popular reporting of AI and robotics, there is a lot of confusion in terms of terminology. There are a couple of papers that back up my perspective though. http://cep.lse.ac.uk/pubs/download/dp1461.pdf takes a historical view on the impact of robotic technology and finds no real difference in the joblessness of industries prone to automation and those that aren't.
http://www.nber.org/papers/w23285 looks at 19 industries that have introduced industrial robotics, and they notice a difference, with each robot equating to around half a dozen humans. The difficulty is that employment figures as a whole (and they looked over like 30 years) didn't change much, but they do note that communities where robot investment was high did struggle to re-enter the workforce.
This has been common with other forms of economic disruption though, with research into former mining communities (https://journals.sagepub.com/doi/abs/10.1177/09500179922118042) finding just 25% of miners had found work a full decade after the mines had shut down! Humans as a species have become so successful in large part because of our adaptability, but there appears pretty strong evidence that in the short-term at least, people are struggling to adapt to losses in their livelihood, with this especially afflicting lower skilled men. Indeed, a report from the government's Foresight division (disclosure, I was a part of that team once upon a time) highlighted the difficulties in getting just this demographic engaged with adult education - https://www.gov.uk/government/uploa...esight-future-of-skills-lifelong-learning.pdf - with both the opportunity and the clear need often insufficient to encourage engagement.
All of which makes it great that the government are, despite the clear warning shot presented by Brexit, doing absolutely bugger all to change things.
Well said.
I actually agree with pretty much everything you are saying on the subject.
From my perspective i was making the point that like you point out confusion has people not realising that there are differences in AI, Automation and overall the use of robotics in general.
So it then makes people think the masses have been and will be let go because of robots and AI
Could there be concerns down the road? Possibly, especially if the people i have consulted for continue to get business, then yes probably.
But its about teaching those what all of it means and also letting those being replaced what options and potential careers they might have the opportunity in.
Thanks for the links also. I have not read any of it before.
Here's the thing I guess. I'm fairly agnostic in terms of the AI apocalypse stuff you hear, but lets assume that the exponential growth guys like Kerzweil talk about will happen, then in that case the economic disruptions of the past are illustrative as they show the adaptability of communities to job disruption. A couple of things are interesting in the data I've seen. As said before, communities are pretty rubbish, that's a given across the world. Secondly, it's been fairly consistent across both regions and type of disruption (economic or technological) that jobs did return in time, but that they nearly always required higher skills than the original jobs did.
You couple that with rising life expectancy, and with that the inevitability that people will be working longer, and the various predictions that a career for life is largely a thing of the past and that people will have multiple careers etc. (kids today entering jobs that don't exist and all that schtick), and the overwhelming sense is that people need to up their adaptability. The wealthy and the highly skilled will be fine as they generally always are, but the poor and the lower skilled are likely to struggle unless they're given an awful lot more help.
The longevity situation uncovers many of the problems,as you'll inevitably get blokes (it's usually blokes as blokes do more manual work) arguing that you won't be fighting fires/lifting bricks/insert other highly physical job as a 70 year old, so therefore the whole thing is bonkers. The very concept of them doing something else for a living is alien, because they've been conditioned to think you have one job your entire life and if that goes, that's your life on the scrap. I'm not at all convinced the government really get it (or care to be honest).
Sorry mate, I meant to reply to this. I think we've come a long way with driverless tech, but that final few percent will be really tough. I mean only this week there was a paper published highlighting the difficulty driverless tech has in distinguishing pedestrians, so it tends to discriminate against non-white pedestrians in terms of safety. The legislative side will take a while, as will modifications to urban planning. One senior Ford bloke I spoke to laughed when I asked him if he expected to see the tech on the roads any time soon.
With other industries, quality of data remains a big issue, with a lot of data science folk complaining that their organisation's data is either all over the place and in whacky silos, or really poor quality that isn't machine readable or usable in any meaningful way.
That's certainly the case with the NHS, and for a couple of years I was trying to get a cross-sector consortium together to do something with health data, much like the Biobanks have to a small extent, and stuff like Project Baseline have been trying to do in the US (albeit as a private project where Verily/Google own all of the data!). It's a huge deal, but the NHS are painful to work with, and I can see the opportunity largely being missed as a result. There is an interesting project in Canada (Toronto I think as a result of the amazing AI dept at the university) where they're looking to bring together medical records, lifestyle data (from wearables etc.) with genomic data, and they can do that because they have a single payer system that theoretically makes it easier. The NHS have that with 60 million people. Huge opportunity and they're gonna blow it by being so slow and disjointed. Apple have already been working with a couple of health insurers, and with all of the tech companies getting involved there's likely to be a big bun fight around ownership of our health data. It certainly shouldn't be the tech companies, and personally I don't think it should be the state either, but I can see it inevitably falling into the hands of a Google as you see with stuff like 23andMe how people happily give control of their data up if they get something even tangentially useful.
Of course, I can see China getting there first because they don't get a monkeys about privacy or any of that, so will have access to a shed load of citizens data.
This is an issue of data, something that - given the appropriate amount of time and consideration - can be rectified fairly quickly. I help teach Intelligent Systems here, and one of the the things I'm keen on doing is talking about how our data reflects our society, and that we need to be conscious of that fact when our algorithms report racist/misogynistic results.
Ontologies, my dude.
My new RSE role involves building a "bespoke" open source biobank management system that'll focus entirely on interoperability and standards-compliance. The goal is to eradicate the scurge of proprietary software from the Welsh NHS - and this is a fairly positive step in that direction. I'm not sure that this is the correct avenue to really detail the project, so feel free to message me if you feel like you can contribute in whatever way you see feasible.
Automation is inevitable, and unless we want to have people employed to take cans of air and spray out dust from arrays of GPUs - then we need to act now.
Rutger Bregman speaks fairly well on this.
The worst thing about my job is the anxiety of waiting for results to come back, our Titan rig went down after Christmas and they've still not got around to sorting out a replacement - so it's all being done on a (yes, one) 1060 with silly small batch sizes (and obviously a silly learning/decay rate).
Black Belt Jones
Player Valuation: £25m
NHS Scotland sitting on a goldmine Bruce. Nowhere else on earth has that many unhealthy people with first world levels of healthcare and corresponding data.
- Status