Will AI take your job? (Dispatches C4)

[Railwaycat]

Sounds an interesting programme (if Radio Times is to be believed), professional people up against AI. Looks like the answer is not yet, but in a few years time might be a different story. Given what's happened today https://www.bbc.co.uk/news/live/c5y8k7k6v1rt (Amazon internet outage) lets hope common sense prevails, although I'm not holding my breath...

• Amazon Web Services says it has fixed the underlying problem that has disrupted many of the world's biggest websites and apps, and that all services have "returned to normal operations"
• Snapchat, Duolingo and Roblox were among those affected earlier, with platform outage monitor Downdetector saying users are now reporting issues with Grok, Lyft and Hulu
• Some applications came back online after more than six hours of disruption, while more than 1,000 companies were affected

 

[Bletchleyite]

This is where people really need to decide what they prioritise. Clearly Amazon and the likes are likely to move to an AI-only customer services operation, and probably Ryanair too. Or if we prioritise better service we'd need to be willing to pay for it.

 

[Railwaycat]

This is where people really need to decide what they prioritise. Clearly Amazon and the likes are likely to move to an AI-only customer services operation, and probably Ryanair too. Or if we prioritise better service we'd need to be willing to pay for it.

But will we be given the choice?

 

[najaB]

But will we be given the choice?

Probably, yes. Companies will move to tiered service where the majority of their customer interactions will be AI-driven but with options for bespoke service for those who are willing to pay more. We already do it to some extent in my company with differentiated support - those who pay for priority support have a separate phone number that routes to more senior techs and benefit from a tighter SLO.

 

[johntea]

It might make some jobs no longer required but then there will be another job to support it

A bit like how my company proudly boasted they would be paperless within a couple of years...15 years later and paper/printers are still very much in demand! But they have replaced *some* processes successfully

Also good luck when something like AWS goes down for a morning, not that things like that ever happen of course...

 

[bleeder4]

Conversely, there are plenty of jobs that AI is creating. After all, AI is maintained and upgraded by human engineers. I recently attended a webinar and the guy running it had the job title "AI Prompt Engineer". His job was to create prompts for companies that they could then paste into AI in order for it to produce the results they were after. That was his full-time job, he did nothing else.

 

[Thirteen]

The AI presenter on the documentary wasn't very convincing but maybe that was the point.

 

[DarloRich]

Surely you still need "human in the loop" in that jobs become AI supervisor/checker

 

[Bantamzen]

I'm very glad that I should be able to hit at least a semi-retirement inside of 5 years. Working as a data engineer in the public sector, I can see that once AI beds into core services, it'll be more than capable of producing it's own data. And whilst there will be a period of transition from writing code to machine learning, eventually I can see my team being largely surplus to requirements in a few years. So my 43 years of paying into retirement and potentially a part time shelfy stacky job down the local supermarket just to top up will be as much as I will be doing (assuming robots aren't stacking shelves by then of course).

But it will be interesting to see just how far AI will reach into the employment market, I suspect it will go a lot further than many think which I think was the point of last night's Dispatches. Once at a certain level, how much will we trust AI to self-govern, as well as undertake tasks and work flows? I suspect the pressure to reduce costs and increase productivity will drive it further than many will be comfortable with.

 

[bleeder4]

I'm very glad that I should be able to hit at least a semi-retirement inside of 5 years. Working as a data engineer in the public sector, I can see that once AI beds into core services, it'll be more than capable of producing it's own data. And whilst there will be a period of transition from writing code to machine learning, eventually I can see my team being largely surplus to requirements in a few years. So my 43 years of paying into retirement and potentially a part time shelfy stacky job down the local supermarket just to top up will be as much as I will be doing (assuming robots aren't stacking shelves by then of course).

But it will be interesting to see just how far AI will reach into the employment market, I suspect it will go a lot further than many think which I think was the point of last night's Dispatches. Once at a certain level, how much will we trust AI to self-govern, as well as undertake tasks and work flows? I suspect the pressure to reduce costs and increase productivity will drive it further than many will be comfortable with.

Yes, agreed. I'm a copywriter by profession and can see myself becoming surplus to requirements within around 5 years. At the moment AI development is being held back by power/energy requirements. AI just needs so much power. Once there is a breakthrough in that, then AI will leap right forwards.

 

[McRhu]

Speaking as an erstwhile artsy, designy individual AI is biting the creative cherry right up to its back molars. It's the tip of a trend that started with DTP, phone photography and social media. I may have alluded before to my early days in design, working for a newspaper that still used hot metal linotype and letterpress. Offset printing arrived and all the traditional caseroom skills were wiped out and I don't think many of the workers kept their jobs. DTP and digital printing then came along and jobs such as rough artist, typesetter and - to an extent - proof reader fell by the wayside. In 1991 was offered a job in a small design company employing 20 people; two years later it was down to one man and his Mac in his livingroom. In later years I worked for different small design studios specialising in design for print. Businesses switching to social media has taken away almost all of that work (at least in my experience). Phone photography took away an awful lot of the photographic work we used to outsource. AI is the most voracious of all the trends to date though. AI writes copy that is often 'good enough'. It produces illustrations, diagrams and photographs that are 'good enough'. And its abilities are increasing all the time, and accelerating at that. In art circles AI art now sells, and is becoming accepted in its own right. It can produce incredible 'work' and it seems that some (many?) customers are now 'doing it for themselves' rather than buying art. I believe it has hit digital artists hardest pro-tem. This is a tide that has yet to sweep all the way in though and I don't even know if there's a high water mark.

 

[Bletchleyite]

Yes, agreed. I'm a copywriter by profession and can see myself becoming surplus to requirements within around 5 years. At the moment AI development is being held back by power/energy requirements. AI just needs so much power. Once there is a breakthrough in that, then AI will leap right forwards.

Certainly writing generic content is likely to become an AI thing, already is to a fair extent. But people do seek expert human analysis too.

I can see it writing all the clickbait for Reach Group certainly (would they need any employees?) - but would it knock out respected writers in the likes of the Times? I doubt it.

 

[DarloRich]

One consideration is the amount of power and data centre space AI needs to operate. It is already pretty vast and will only get bigger.

 

[cb a1]

For anyone interested in the more philosophical elements of this topic, I would highly recommend reading Nexus by Yuval Noah Harari.

There's a big difference between generating information and making decisions. Using AI for generating information from data is largely an extension of what we've been doing since humans started to form civilisations and it can lie just as well as any human can (as shown by millions of videos and clips on social media).

When it comes to decision-making, you have to start considering how risky and thus how accountable that decision needs to be.
E.g. Letting AI decide when to switch your heating on is a low level risk.
Letting AI decide whether to respond to what it thinks is a nuclear strike is a high level risk (as we saw in the 1983 film Wargames).

The one big take-away I got from Nexus was to always think of AI as Alien Intelligence rather than Artificial Intelligence. It can pretend to be human and in many situations can fool people into thinking it thinks like a human, but ultimately it's an alien.

Sci-fi has been exploring this concept for decades now. I'm not much of a trekkie, but I've seen enough to know that Spock and Data were used to explore this tension to great effect with each trying to understand 'what it is to be human'.

 

[liamf656]

The AI presenter on the documentary wasn't very convincing but maybe that was the point.

Not now but if we see how much it's progressed it definitely will be convincing soon

Surely you still need "human in the loop" in that jobs become AI supervisor/checker

I can see this being the case, however (for arguments sake) it would probably be a handful of people overseeing a job that used to be done by thousands of humans, but it's just my opinion

Yes, agreed. I'm a copywriter by profession and can see myself becoming surplus to requirements within around 5 years. At the moment AI development is being held back by power/energy requirements. AI just needs so much power. Once there is a breakthrough in that, then AI will leap right forwards.

I can see this happening for sure

A few jobs I believe will be taken over by AI in the distant future are
- Translators and interpreters
- Editors and writers
- Presenters and Radio DJs
- Telemarketers
- Call handlers and switchboard operators
- Public information (transport and tourism)

And the jobs I consider to be safe in the long run seem to be
- Manual handlers
- Labourers (bricklayers, decorating)
- Mechanics
- Heavy machine operators (less so)
- Construction workers
- Catering
- Anything involving boats
The lists are non-exhaustive but I think everyone in every role should consider how artificial intelligence could affect their future. Even cars are driving themselves, food is being delivered to people in a box that transports itself and so on. It is rapidly advancing and as mentioned, once the power is obtainable it will continue to grow at such a rate

 

[styles]

I sit in a bit of a funny position on this.

My main bag is software engineering. However much software engineering I get up to isn't particularly difficult. Yes there's some algorithmic work to do, but there's also a lot of writing simple queries. The latter can be undertaken by AI solutions today. In fact in my development environment I have AI tools which write this code for me and I simply review it. I rarely find the need to change it, so to an extent, AI is already doing part of my job.

I do feel that to this extent, I should upskill to ML and AI software engineering. I have enough other strings to my bow (architecture, consultancy, disaster recovery, etc) that I'm not at immediate threat of being replaced by AI, but I'd be surprised if the work I'm doing today isn't majority automated in 5 years' time.

Mind you, I'm increasingly of the opinion that sitting in front of a computer for 8 hours a day isn't healthy, mentally or physically, and I should transition into outdoor leadership, which isn't really at risk of AI replacement unless you're confident having a robot belaying you on a multi-pitch climb.

 

[Bantamzen]

Not now but if we see how much it's progressed it definitely will be convincing soon


I can see this being the case, however (for arguments sake) it would probably be a handful of people overseeing a job that used to be done by thousands of humans, but it's just my opinion


I can see this happening for sure

A few jobs I believe will be taken over by AI in the distant future are
- Translators and interpreters
- Editors and writers
- Presenters and Radio DJs
- Telemarketers
- Call handlers and switchboard operators
- Public information (transport and tourism)

And the jobs I consider to be safe in the long run seem to be
- Manual handlers
- Labourers (bricklayers, decorating)
- Mechanics
- Heavy machine operators (less so)
- Construction workers
- Catering
- Anything involving boats
The lists are non-exhaustive but I think everyone in every role should consider how artificial intelligence could affect their future. Even cars are driving themselves, food is being delivered to people in a box that transports itself and so on. It is rapidly advancing and as mentioned, once the power is obtainable it will continue to grow at such a rate

I think the Dispatches edition in question also demonstrated that more traditionally "safe" office work could come under threat such as solicitors, middle management, consultants etc etc. These jobs will be at particular risk as they sometimes add a lot of cost to not a whole lot of value, and some facets of their work could be handed over to AI to be done at a fraction of the cost, as demonstrated by the solicitor's case on the show.

It's something I recently pointed out on a work call to middle managers excitedly talking about AI in our workplace. I told them lot of the kinds of decisions they make could easily be done by AI, and without the need for endless Teams calls & PowerPoint presentations, leading to a stunned silence! We have to remember AI isn't just automating set processes, it is applying Machine Learning to them to streamline & improve processes, and allow AI to take on decision making around these changes. I think a lot of people could be in for a big surprise in the next 10 years.

I sit in a bit of a funny position on this.

My main bag is software engineering. However much software engineering I get up to isn't particularly difficult. Yes there's some algorithmic work to do, but there's also a lot of writing simple queries. The latter can be undertaken by AI solutions today. In fact in my development environment I have AI tools which write this code for me and I simply review it. I rarely find the need to change it, so to an extent, AI is already doing part of my job.

I do feel that to this extent, I should upskill to ML and AI software engineering. I have enough other strings to my bow (architecture, consultancy, disaster recovery, etc) that I'm not at immediate threat of being replaced by AI, but I'd be surprised if the work I'm doing today isn't majority automated in 5 years' time.

Mind you, I'm increasingly of the opinion that sitting in front of a computer for 8 hours a day isn't healthy, mentally or physically, and I should transition into outdoor leadership, which isn't really at risk of AI replacement unless you're confident having a robot belaying you on a multi-pitch climb.

Oh you are absolutely right here, if your job (like mine) is often just predicated around writing code for querying data, then it's time to upskill or plan a new career. Personally I'm close enough to retirement not to be overly concerned, if my job goes from under my feet just pay me off and I'll walk. But for people planning to stay in the industry for a long time, upskilling to Machine Learning would be very prudent. It'll still be some time before AI becomes good enough not to require human interventions at the highest levels, and in fact ML jobs are likely to be the ones that attract the really big bucks

 

[ls2270]

Not now but if we see how much it's progressed it definitely will be convincing soon


I can see this being the case, however (for arguments sake) it would probably be a handful of people overseeing a job that used to be done by thousands of humans, but it's just my opinion


I can see this happening for sure

A few jobs I believe will be taken over by AI in the distant future are
- Translators and interpreters
- Editors and writers
- Presenters and Radio DJs
- Telemarketers
- Call handlers and switchboard operators
- Public information (transport and tourism)

And the jobs I consider to be safe in the long run seem to be
- Manual handlers
- Labourers (bricklayers, decorating)
- Mechanics
- Heavy machine operators (less so)
- Construction workers
- Catering
- Anything involving boats
The lists are non-exhaustive but I think everyone in every role should consider how artificial intelligence could affect their future. Even cars are driving themselves, food is being delivered to people in a box that transports itself and so on. It is rapidly advancing and as mentioned, once the power is obtainable it will continue to grow at such a rate

Other jobs I would add to the ‘safe’ list are plumbers, electricians, hairdressers/barbers and beauty services like nail technicians, make up artists etc!

 

[PTR 444]

Other jobs I would add to the ‘safe’ list are plumbers, electricians, hairdressers/barbers and beauty services like nail technicians, make up artists etc!

Also jobs where humans are required by law for safety purposes, such as cabin crew.

 

[styles]

Also jobs where humans are required by law for safety purposes, such as cabin crew.

I'm sure we once said this about rail, but driverless trains are very much a thing now, even if not in the UK so much.

 

[PTR 444]

I'm sure we once said this about rail, but driverless trains are very much a thing now, even if not in the UK so much.

Yes, but even the DLR still requires a human on board for emergencies.

 

[styles]

Yes, but even the DLR still requires a human on board for emergencies.

Yes, the UK is quite behind some other countries on the driverless train front. But driverless trains do exist around the world, as do driverless taxis (coming to London soon). It's not inconceivable to imagine a point in the future where we won't need airline pilots. In fact some countries are trialling unmanned drones for deliveries already, including Orkney soon. Perhaps a central remote controller role may remain for an intermediate period, but a pilot for each plane probably won't be necessary in the future.

 

[brad465]

If AI causes a notable net loss of jobs, this may not be a problem as much of the developed world sees fertility rates well below replacement rate, so in time a natural population shrinkage happens anyway. However, what will be a problem is if the rise of AI and the decline of population do not align well, which is very hard to coordinate even if we tried to align them. The youngest baby boomers are in their early 60s, we probably need another 20 years before they've mostly gone and society can cope with fewer jobs paying the pension bill.

 

[Bletchleyite]

Yes, the UK is quite behind some other countries on the driverless train front. But driverless trains do exist around the world, as do driverless taxis (coming to London soon). It's not inconceivable to imagine a point in the future where we won't need airline pilots. In fact some countries are trialling unmanned drones for deliveries already, including Orkney soon. Perhaps a central remote controller role may remain for an intermediate period, but a pilot for each plane probably won't be necessary in the future.

The main issue with totally unstaffed trains is poor customer behaviour. But "guard only operation" is likely to become feasible outside of metro networks at some point (certainly on the likes of HS2 where there are no level crossings etc), and guards are a LOT cheaper than drivers.

Isn't the Glasgow Subway meant to be going to completely unstaffed operation once the platform edge doors are all installed? Presumably security guards (generic ones as you'd use anywhere else) will be used to deal with problem times of day for antisocial behaviour.

 

[PTR 444]

Yes, the UK is quite behind some other countries on the driverless train front. But driverless trains do exist around the world, as do driverless taxis (coming to London soon). It's not inconceivable to imagine a point in the future where we won't need airline pilots. In fact some countries are trialling unmanned drones for deliveries already, including Orkney soon. Perhaps a central remote controller role may remain for an intermediate period, but a pilot for each plane probably won't be necessary in the future.

Even if we got to a point where no airline pilots were required, I think that a lot of passengers would feel uncomfortable flying in a fully staffless plane. My prediction is that there will still be a requirement for cabin crew to be on board, who can take over control of the plane in an emergency.

 

[Bletchleyite]

Even if we got to a point where no airline pilots were required, I think that a lot of passengers would feel uncomfortable flying in a fully staffless plane. My prediction is that there will still be a requirement for cabin crew to be on board, who can take over control of the plane in an emergency.

As with long distance trains I can't imagine there being no sort of service staff on board an aircraft. And you do need to manage fast evacuation. However almost all air crashes are caused by human error, so even if it would make some types of defect less survivable safety would be improved significantly overall by removing pilots once this is technically feasible, though risks posed by hacking of any remote control feature would also need to be considered. This is already looked at in a way by the Airbus flight laws which won't let the pilot stall the aircraft in normal operations - Sullenberger I think was quoted as saying he couldn't have done the "miracle on the Hudson" in a Boeing as he'd have had to worry about stalling, whereas in an Airbus he just pulled hard back on the stick and the aircraft gave him the longest glide it could safely achieve without stalling. It might have been that the Boeing could get slightly further, but it would have required an incredibly skilled pilot to achieve it without stalling and crashing at such a low altitude.

As for trains I have no qualms about the idea of a fully driverless train; the signalling system is what dictates what is safe, and having the bit of squishy meat between the signalling system and the controls has been necessary up to now because of limits to technology, but when it no longer is the system will be that little bit safer by removing them from the pointy bit and putting them in the passenger cabin to deal with people.

But is that AI? I'd argue not - it's just simple automatic operation by making the train do what the signalling tells it to without a person in the middle looking at lights on sticks. The only applicability of AI might be in obstacle identification, but to be honest you can just do that with radar or similar.

AI might however be able to enable safer automatic door control by being able to identify obstructions to doors on a busy platform while not making it possible to just hold the doors open for as long as you like as you can with a lift.

 

[najaB]

Even if we got to a point where no airline pilots were required, I think that a lot of passengers would feel uncomfortable flying in a fully staffless plane. My prediction is that there will still be a requirement for cabin crew to be on board, who can take over control of the plane in an emergency.

It will be a *long* time before completely autonomous passenger aircraft are common.

If the AI in a driverless train has a problem, it stops, "signals" are turned to red and everyone is safe. If the AI in a driverless car has an issue, it pulls to the side of the road, other vehicles either stop or drive around it, and everyone is safe. If the AI in a pilotless airliner has an issue there's a smoking hole in the ground.

And the "cabin crew who can take over in an emergency" doesn't take into account how quickly those emergencies can develop. Just look at the numerous examples we have today of pilots who are already sitting in the seat have been caught out when the automation kicks off and hands them the aircraft in an unexpected state.

 

[styles]

640K RAM ought to be enough for anybody

 

[Bletchleyite]

And the "cabin crew who can take over in an emergency" doesn't take into account how quickly those emergencies can develop. Just look at the numerous examples we have today of pilots who are already sitting in the seat have been caught out when the automation kicks off and hands them the aircraft in an unexpected state.

That of course is automation designed for the presence of a human who can take over. Automation designed for a pilotless aircraft could not work in that way.

Even so, I'm fairly convinced that stories of pilots having caused a crash (or a near crash which they then averted) are likely to be far more common than stories of pilots heroically saving an aircraft. But that's hard for people to accept psychologically, just like it is to accept that driverless cars will be safer overall by making some types of crash less likely but some more likely. (Though that is accepted in road design - I'm sure I recall reading it was an accepted aspect of the Hemel "magic roundabout' that it would cause more minor damage only accidents than before but would reduce the number of killed/seriously injured ones substantially).

 

[The Pelican]

It's always seemed a funny thing to me that the automation of transport modes has followed a trajectory from the most complex types of movement to least complex- the Mecury Capsule already had an autopilot in the 1960s, as did the Victoria line in 1967, plane autopilots came in soon after that, but car automation is still a long way off, and mainline train automation behind that. The trajectory suggests that an automatic horse control system will roll out in 2050.