But this is exactly the right reading of the chart - the description says "chances that automation will lead to job losses", not "chance that all jobs will be replaced by automation". So, perhaps it will be drones assisting in wildfires that reduce the need to pay as many people to fight that type of fire, whilst structural firefighting remains the domain of humans - for now.

Why not? Is it the chaotic environment? The temperature extremes? The unpredictability of each situation?

All those are strong reasons TO automate firefighting. Machines can function better in heat/smoke/violent structural collapse situations. They can make objective analysis when their own mortality is not an issue.

Nobody made money betting against advances in computing.

I don't think that the general idea is replacing specific jobs with robots. Probably not very pragmatic most of the time. Rather it's jobs that simply become obsolete due to technological advances.

You could probably eliminate the need for most firefighters by installing more and smarter networked sprinkler systems, using different materials for building, building smarter oven's / toasters / whatevers, or any number of other innovations... but all of that costs. It seems less threatening when viewed this way, instead of a robot replacement, but the effect is the same.

Automation is also elitist. I'm betting that a building housing Google's servers has more fire prevention capabilities and far less need of human firefighters than say a sweat shop in Bangladesh.

But to replace a fire fighter would require order of magnitude discoveries in several fields you would effectively need true AI and a self powered robot with as good strength and endurance as a human.

Don't assume that the robotic solution will look anything like the human solution. A fitbit does not look like a "robotic personal trainer" and a roomba does not look like a person with a broom.

I would speculate that the obvious next wins for fire-fighting are drones to visualise the scene from multiple angles, remote control or self-aiming hoses and extinguishers. And various hardened eyes and arms to reach into the fire. All with human oversight, but a reduction in the number of humans involved or in harm's way.

This reduction in manpower is the pattern - e.g. supermarket self-checkouts still have some human staff, but to troubleshoot and supervise.

I'm actually most of the way done with a prototype of a quadcopter that can be deployed on a scene that just orbits and streams video to a monitor in the chiefs vehicle. It's mostly off the shelf components, with a little bit of 'wrapper' that makes is easier to deploy.

The biggest issue I haven't sorted out yet is the batteries... LiPo batteries are great in terms of capacity/weight, but if you keep them fully charged all the time, you kill them pretty quickly...

Swarm them.

When one is approaching 20%, send one that's charged to replace it. Once the replacement has arrived, return the depleted one to the charging station and do not charge past 80% or so.

If you make the charger smart enough to not completely charge the battery, and you have a ..eh.. Battery of copters available, you've solved the problem

Could they be stored at a low temperature until there's an actual need to use the copter?

Ok can this go into a burning building find trapped people and carry them out?

No? That's well outside the scope of its design... what's your point?

Right a robot that can perform as well as a fit human firefighter will be available any time now just like fusion, Hard AI is only 5 years away and have been for the last 50 years.

And for that matter, better fireproofing.

Those things are improving every week. Self-powered robots will exceed human strength and endurance in a few years with steadily improving battery power/kg (metal-air batteries are already better than gas-powered engines). AI for autonomous robots is the current Darpa challenge. And for firefighting, the automation doesn't have to extend as far as life-or-death decisions; those can be made remotely.

Though I'm not sure it needs to be completely automated. You don't see robots writing news articles, but a researcher and Google are more productive than several researchers.

I'm all for advances in technology making things safer and putting fewer people in harms way. I just don't think there is anything close to a 17% chance that it will happen in the next two decades. I say this as someone who has a foot in both worlds, so to speak... I work in technology, build UAV quadcopters (including playing with various sensors) as a hobbyist, and work at least one shift a week as a firefighter/paramedic.

Technological advances in recent years have made life much better in the fire service. Thermal imaging cameras have gotten lighter, with much larger thermal ranges. Gas detection has similarly lighter and more accurate. Breathing apparatus have gotten 'smarter' and lighter, allowing longer working times, less fatigue, and better insights for incident commanders on the status of their crews.

Research in fire development and behavior has led to significants advances in strategy and tactics. However, those tactical decisions are still well beyond the scope of what computers (and computer vision) are capable of. There is just too much variation in how homes are build and occupied (and modified).

Here are just a few of the issues I see:

-Moving an unconscious victim requires a fair amount of care, and I don't think robotics are there yet...

-'Reading' smoke coming from a structure gives a lot of insight into the design and occupancy of the structure. I think that's well beyond the current limits of computer vision and decision making.

-There are many activities performed by firefighters outside the scope of 'putting water on fire'. Responding to medical emergencies, technical rescue scenarios, etc... Extricating a victim of a car crash is a very delicate operation that employs equipment capable of exerting many thousands of pounds of force, and a few inches is the difference between freeing someone from a deformed care and crushing them to death...

Really the biggest issues I see have to do with tactical decision making based on a huge number of subtle variables (which would involve very sophisticated computer vision), and simply navigating in and around a structure. Designing a robot that can move autonomously through all areas of a house, into closets, under beds, etc, while still being strong enough to lift and move a disabled victim, or employee various tools seems like a pretty difficult problem to solve. How would you power it? Batteries aren't anywhere close to that point, and tethers would limit mobility too much.

Personally I think the next 20 years will see continued augmentation, rather than replacement of the men and women in the fire service. For instance, I would love to see an SCBA facepiece with an integrated heads up display providing an infrared image and levels of various gasses that are good indicators of air quality. Once we have that, I'd love the ability to switch between 'tank' air (which I carry a very limited amount of on my back) and 'filtered' air (which would just come from the environment, filtered for particulates, allowing me to save my bottle for when I need it).

I think there is also an opportunity for improved 'mapping' of structures as a firefighter moves through them (again, integrated on a HUD inside the facepiece). That would significantly reduce the number of deaths and injuries that occur when firefighters become disoriented and lost in zero-visibility situations.

This ended up being much longer than I expected it to be... TL;DR; The next 20 years will see augmentation, not replacement. Who knows what the 20 years after that will bring...

It seems like you're stuck on the automaton aspect. Would not remote control robots decrease the number of firefighters necessary? If each building came equipped with remotely operated firefighting drones, response time would improve by orders of magnitude.

The same for car accident incidents: would a precisely controlled remote robot not be able to more aptly apply the jaws of life, especially in situations dangerous for a first responder?

Of course there are a bevy of situations where first responders are still absolutely necessary. But other things also work in tandem to improve the outcome: like civilian operated defibrillators. One could also interpret the 17% as "17% more effective, through automation, per person employed." This would, as a ratio of field employed to population, look less like jobs lost and more like additional jobs unnecessary.

>Would not remote control robots decrease the number of firefighters necessary?

Perhaps, but I think the loss of situation awareness would be a big problem to overcome (certainly not insurmountable, just the first challenge I see).

I absolutely agree that this is the direction things are headed, I just think the timescale is a little optimistic.

Sounds like you have a good grasp of the subject. One comment: as for batteries, they are almost there, not at all far away. Metal-air batteries are already capable of over a thousand watt-hours/kg.

As for reading smoke, a machine could be adapted to ready everything - heat, smoke, particle size, explosive residue, gas leak - and factor all that in. I don't mean to denigrate an experience firefighter's skills. But a machine can sense far beyond what eyes and nose can do.

The augmentation idea is very cool. Sensors are clearly an advantage. And soon (sooner than 20 years) the suit will be more powerful than the human. Hands as powerful as jaws of life! A back that can support a roof!

I don't know much about metal-air batteries... Based on some quick reading, it seems most of them use water as the electrolyte solvent. Wouldn't they have issues with ambient temperatures north of 100˚C?

water doesnt boil in sufficiently small sealed containers-- the pressure keeps it liquid.

Perhaps I'm misunderstanding how metal-air batteries work... I thought they used ambient air as the oxidizer, precluding them from being sealed.

I was under the impression that more modern building construction and heating technologies have already made a large percentage of professional firefighters redundant - hence why they spend as much time responding to car accidents and working as EMTs as actually fighting fires.

The number of fires has indeed fallen dramatically. However, until the day when there is zero risk of there being a house fire (and we're a _long_ way from that), you still need to maintain a staffing level that can get the job done.

Last year, London announced plans to let go 10% of its firefighting force. This is despite booming population and construction. In 2004, there were 572,000 fires in the UK. In 2013, there were 193,000 reported fires. (https://www.gov.uk/government/uploads/system/uploads/attachm...)

Sure, when you scale it up to the something the size of a city, then you can say "We want to be able to handle X major incidents at the same time over the entirety of the city."

My point is that X will continue to shrink (as has been the trend for decades), but never hit 0, and you can only consider a limited area before response times become an issue (even if you somehow got to the place where you were very confident there would never be more than one concurrent fire in the UK, obviously you couldn't reduce staffing to a single company).

Actually, that 17% doesn't mean that robots will replace firefighters -- it just means there will be 17% less need for human firefighters. This could take a number of forms -- we've already seen some of them, such as better electrical codes that have reduced the number of electrical fires.

Another recent trend is residential sprinkler systems. Now imagine a smart sprinkler, with thermal imaging cameras that can pinpoint the source of a fire before it spreads. Then activate a directed jet spray to that source, and you have a whole building fire that has been avoided.

Or, lets say if 17% of fires are caused by space heaters. What if the heaters were equipped with a thermal imaging sensor, so that it cuts power to itself if an object in front of it gets too close to ignition temperature?

Hmm... upon closer reading, I see the charts says that there is a 17% chance that there will be some degree of job losses due to computerization. I think that's a more reasonable thesis (though not for the reasons you mention).

Fire prevention will only take you so far. You still need to be able to deal with the inevitable fires that still occur, and you need to have the capacity to deal with them 24/7/365. You can only reduce staffing so much and still have the ability to put a single structure fire out, and many departments are at that point already.

I do see a potential for job losses due to augmentation technologies (exoskeltons, etc) that will enable one firefighter to do tasks that previously took two, etc...