At almost every partner site I bring my robots to, I observe something similar. Workers on the factory floor look at our robots curiously and mumble to each other as we bring them in. Their eyes widen when we do a demonstration of our robots doing a simple everyday task they usually do. Sometimes if the demonstration does something that workers don’t like to do, they applaud. If it’s something that is squarely within their job scope, usually they look at each other nervously.
“They’re thinking whether robots will replace them,” one operations manager (Mr. O) remarked to me. “I think it’s good that robots do. They will have to upskill.”
I’ve spent a lot of time thinking about robots and their implications on society more broadly. I am bullish on robots and believe there will be significant societal implications if robots become ubiquitous.
Let’s explore some possibilities.
Possibility 1: Robots will replace our jobs, which will lead to a dystopian future of mass unemployment.
I won’t spend too much time on this as this argument often comes from an incomplete understanding of how technology is developed and adapted, and thereafter deployed in society. It also fails to account for the fact that there will be some form of transition, rather than an instantaneous replacement of workers.
A more nuanced possibility is as follows.
Possibility 2: Robots will replace jobs, humans will need to upskill.
Mr. O’s argument follows this logic. It’s not wrong, though it should be said that the logic holds to almost any type of technology or automation and is not unique to robotics. There is no doubt that this will happen and is happening in the robotics space. We have seen this already happening with autonomous driving, manufacturing, logistics and more, with Amazon’s warehouses being a prime example.
Yet this possibility misses two things.
First, it takes the perspective that robotics is like any other technology or automation - which is fair perspective to have, and remains to be seen what happens - yet I am more bullish than that. I think robotics as an industry will be bigger than the internet. I believe robots will radically reshape the way work is conceived; and also am certain all sorts of laws, ranging from tort to IP, will need to adapt and be passed with this new technology.
Second, it neglects important points of whom, how and when. Put differently, it glosses over the difficult questions of whose, what, how and when jobs are to be replaced and upskilling happens. There will be significant teething issues that cannot be glossed over by “humans will adapt”. We, as humankind, may, but many individuals, families, and communities might not.
While sound, this possibility is not nearly enough to capture the nuance of a future with robots.
Possibility 3: Robots will liberate humans from work, freeing up time for leisure and for human flourishing.
You might have seen this argument before. Elon Musk said that humanoid robots will elimate poverty. I’ve seen tons of robot start-up decks’ and website’s making the same claim.
Like Possibility 1, this possibility fails to account for how complex it is for robots to be developed and adopted in society, though it suggests that society does adapt to the transition to robot work by redistributing time and wealth to humans. It may be worthwhile to unpick this a little more.
This argument is premised on the fact that (1) there is widespread adoption of robots, (2) there are more robot workers (perhaps supplanted by some human workers) than there are jobs, and (3) society develops to a point where humans can pursue leisure and human flourishing without the need for income via jobs.
I cannot currently imagine the widespread adoption of robots to the point where there are more robot workers than there are jobs. It is an incredibly difficult technical challenge to have robots perform any and all tasks - even simple tasks like grasping with the right force, taking things from dark spaces (occlusions), and so much more. For the sake of argument, let’s assume this premise that this technical challenge is surmounted, perhaps with enough data, compute and better models. I still think this is unlikely.
Let’s put aside nuance and differences in types of work for now. It is important to note that robots have a floor price - there are costs involved in the hardware, software, AI models, compute, data and more. Even with scale, such a floor price still exists. There will be a point where the unit economics of robots do not make sense (even after factoring in robots working 24/7). I’ve explored this extensively when thinking about the kinds of partners to deploy robots in. There are many, many industries where the unit economics are hard to justify, and I think it is unlikely robots will replace human labour entirely.
We can then get into jobs where it is not just difficult, but makes less sense for robots to replace humans, for example industries like healthcare, social care, etc.
This argument also misses the whose, what, how and when of the transition that I highlighted above in Possibility 2. Further, it also misses the why - why would the wealthy (the haves) willingly give away significant amounts of his wealth to get rid of poverty, free up time for others, and contribute to human flourishing of the have-nots? While I hope that happens, it’s a tricky future to hope for. Unfortunately, abundance does not imply leisure.
Possibilty 4: Robots will radically redefine work. And, depending on the way we develop robots, humans may no longer need to do dull, dirty and dangerous tasks.
The Vatican mentioned that the greatest liberator for women in the past century was a washing machine. While arguably not a robot, the washing machine brought autonomy to clothes washing and therefore unlocked significant amounts of time for women in the 20th century. There were jobs that were replaced, and humans upskilled. I think they upskilled well, too, with labour participation, education, and work across different sectors.
Robots will undoubtedly do a similar thing that washing machines did (Possibility 2 above). Yet there are more similarities. Like how washing machines replaced a specific type of work, I foresee a future where robots replace many specific types of work, in a piecemeal, possibly sequential way. This therefore depends significantly on the type of work that gets replaced.
I believe work that will be replaced will fall within the “dirty, dangerous, demeaning” paradigm (based on the Japanese expression). I think “dazzling” is one that can be added to the mix - i.e. robots that are used for marketing and the wow factor (think robot bartenders). This is because, as I made the argument above, there ultimately is a floor on costs of robots, and there will be a point where unit economics do not make any sense. The kinds of work that will therefore be replaced are work where it makes economic sense for robots to replace. The idealist in me thinks “dirty, dangerous, demeaning” and “dazzling” work, once we factor in the value of not having humans do such work even if affordable (insect farms? underwater ship repairs? the list goes on).
There may also be a bucket where easy or easily automatable work gets replaced by robots too. We see this in large language models now. Finance or legal work isn’t dirty, dangerous or demeaning, but is getting increasingly automatable by LLMs. This argument holds, yet robotics has a much higher floor than LLMs on costs, so there is a little more consideration required here. Just because something is easily automatable doesn’t mean it will be replaced once unit economics are considered.
I don’t think many people are thinking too hard or critically about this issue right now, as this is still a future that seems quite distant. Fair. Yet, similar to what I raised in my thesis on robot learning in my robotics post, I think careful consideration of what the future may or should be allows us to build more wisely today.
I have a different take from Mr. O. Those nervous looks at my partner sites are not just a reaction to my robots today. They reflect curiousity - neither necessarily optimistic or pessimistic - to the choices we are making now about the kinds of work we decide robots do tomorrow.
I hope to be part of building that tomorrow. It’s exciting.
DNMZ