Often, the most revolutionary innovation arrives in a form that discourages or misdirects the imagination. Take for example the transistor. In 1947, an ugly and unimpressive contraption which excited only a small group of scientists and fans of science, it ultimately had an impact on the course of our technological evolution, and in turn on our society, that no one could have had imagined. Even today, the modest transistor is out of sight but still at work, present in every electronic device. One unit of the newest kid on the block, Apple’s M1 Max microprocessor, has 57 billions of them. 

This analytical essay is about one innovation which I believe it will have that kind of impact. I arrived to this conclusion in a zig zag fashion, no single “aha” moment, just observations and ideas bubbling along until I started to see something different. 

It all began with a unremarkable publicity stunt:

Introducing Tesla Bot

Tesla AI Day 2021, August 19: one hour into the presentation, the last slide draws the final conclusion showing how the next generation of Tesla Dojo supercomputer, soon to be released, will outperform the current version by a factor of 10x, a version that already outperforms anything else out there today by a good margin. Then something unexpected occurred. A strangely dressed man stepped onto the presentation floor and started dancing with the elegance of a damaged pull string puppet until Elon Musk asked him to stop. It was like a scene from The Wiggles show. For sure, everyone in the audience was wondering WTF was that all about. Equally unexpected was the answer: that was how a Tesla robot (named… Tesla Bot) to be released next year will look like, Elon explains. Details followed. These robots are benign humanoids, of little strength and although designed to have impressive kinetic abilities, they are not nearly as good as humans. It sounded like a story AI engineers love to hear. 

Initially I concurred with the idea circulated in the media that this is a marketing ploy to recruit talent, but the more I thought about this robot, the more I began to see this innovation may be not just a Boston Dynamics style concept, but a real manufacturing product with profound impact on our society. Granted, quite a few assumptions need to be validated along the way, but there is merit in exploring these possibilities. We are at an interesting point in our history, when with the passing of each day we seem closer to saying goodby to inventions that carried us through the last century to new heights, such as ICE automobiles, fossil energy, petrol based airplanes, centric, hierarchical business models, mass education, and physical labour. It is all happening, like in a carefully choreographed set. We haven’t been in such moment since the industrial revolution, when everything changed simultaneously. Internet has been a huge change, but as a singular phenomenon. The other parts of the society continued to function as they have been for a long time, in a gradual, cascading advance. Now, everything is in a radical change, including the labour itself. Yes, the points that I am bringing here may look exaggerated, but believe we are not that far.

I looked at the potential impact of Tesla Bot from the perspective of labour economics, its technological impact on productivity, the potential quantum leap in Tesla’s manufacturing competitiveness, the spread of humanoid robots on Earth, the challenges of human settlement on Mars and some aspect of the future of society and work in general. These topics may seem a little stretched and far apart, but it will all makes sense when you see how Tesla Bot is a key component of each.

Economy and Labour

What is economy? Economy is labour, Elon says.  A large subset of that labour is physical, repetitive. In today’s world this type of work is a chore that doesn’t get too many likes. Most of the time is unpleasant, unhealthy, dangerous and not well paid. Physical labour should be optional, he says. The future world is a world where people have a basic universal income and are free to make that choice. We are not there yet, but Tesla Bot will help by taking up on some of that workload, so people can find time to focus on non-physical creative projects, unless they prefer differently.

In the context of macro economy, labour sounds abstract, amorphous, but in concrete situations that perception changes completely. It is like going from vapour to sharp metal. 

Take SpaceX as an example:

On Sunday, February 23, 2020, at 1am Elon Musk called an all-hands meeting at the site where Starship was being built in South Texas. He asked one question: why was the factory not working full throttle, as fast as he wanted? The answer that came back: not enough people.  He then asked the present team members to contact people in their close social network, family and friends, and hire them to work for SpaceX, selectively. They must consider only those who can do the job well and who wouldn’t destroy their reputation. They would be responsible for the person they hire. In less than two days that site doubled its workforce and the work issue was solved.

In this context the issue is a human resources problem,  specific and local, with consequences on the business plan. From this vantage point you cannot extrapolate it at macroeconomic level to conclude this is a wide spread issue. The solution was a temporary quick fix. Temporary, because people move, they tire of the job they did not plan to have, and they may want or need to do other things. In the long run, if applied repeatedly, this quick fix becomes a costly solution.

You may think the staff shortage at the South Texas SpaceX building site is a situation caused by a demanding boss who wanted the work to be done NOW, but before you do that, take a look at the business plan. If the idea is to deliver something by the end of the following weekend the situation is not a labour problem, it is a human resources problem. But if the supply deficit persists affecting the ability to execute a multi-year project plan because there are not enough people for hire, that IS a labour problem. This may sound like blah, blah, blah unless you read the plan: Elon Musk wants to build a production line for Starships that is capable of pumping a Starship every 72 hours in order to put together a fleet of 1,000 ships necessary to populate Mars. How do you execute the multi-planetary expansion plan if you don’t have access to adequate labour? On the background of supply chain problems and hiring difficulties, I have come to believe that Tesla Bot is meant to address this issue, at least partially, in anticipation (which, if realised, it will be a stroke of genius by Elon Musk) of significant labour shortages that will affect developed countries so keen to build a reliable supply chain at home. 

Internal Labour Factors

Executing a plan of this magnitude requires the right skills, the right teams with the right people.  This is a massive start-up that is simultaneously a manufacturing operation, creative engineering, science and arts all in one continuous and coherent flow. SpaceX needs teams that work well together, discover, invent and develop with speed and effectiveness. New disciplines will emerge during this time. For that to happen you need the physical labour in place to complete tasks with well-known methods  so that the innovation teams can focus on finding solutions for complex problems by inventing methods that are unknown at the current stage. SpaceX’s Mars project needs to deliver in a tight time frame and labour can make or break its execution. 

This type of advanced and unique work imposes demands that cannot be satisfied by standard labour infrastructure trained to service typical industrial enterprises.

Tesla had similar problems. Tesla Model 3 production issues in 2018 are well documented. The “production hell” was definitely a stressful experience for all employees. Before starting the Model 3 production Tesla employed an ergonomics team to address the repetitive labour issues affecting workers which occurred frequently during the Model X production. However, when the production of the new model started the physical labour issues persisted and multiplied. Reports appeared in the media describing the difficult working conditions and problems caused by long hours (Elon Musk submitted himself to a an intense working regime too). “Everything feels like future but us”, said one of the workers in an interview quoted by The Guardian

The dependency on human labour in key manufacturing areas heightens the risk of injuries and drop in productivity: “you really can’t have people in the production line itself. Otherwise you’ll automatically drop to people speed”. “People’s role should shift to maintaining machines, upgrading them and dealing with anomalies”. 

These two episodes (at SpaceX South Texas, Texas site and Fremont, California plant) mostly reflect issues related to internal factors. These are planning issues that in general could be prevented with better management. There are attenuating circumstances for both Tesla and SpaceX as they are pioneers in their field and lack the long manufacturing experience of the established auto and space hardware makers (although there is an upside to that inexperience). Reflecting on the experience of building Tesla manufacturing plant, Elon Musk said “this is the worst way to earn money, honestly [starting a manufacturing car from scratch], maybe rockets is a bit worse”. Over time Tesla did improve its auto manufacturing processes to a level where they can be comfortably compared with the traditional incumbents. In fact they are now the best in EV manufacturing. Yet, the human labour shortages are still looming as a large problem now that Tesla is in the course of opening and building new factories. As for SpaceX, that is a completely different case. With no other real-life model to emulate, virtually creating a new industry from scratch, an unforgiving deadline and extraordinary targets, labour shortages present a red-hot, critical risk. 

External Labour Factors

External factors affecting labour are more difficult to predict, especially in new industries. Risk management plans may work in the context of traditional labour market issues (employment levels, wages, GDP growth, etc.), but not when you build a fleet with destination Mars for the first time in history. You need to anticipate that you will have issues that are unsolvable with orthodox methods and use disruptive innovation as a way of preparation. The creation of Tesla Bots looks like that kind of anticipation. 

An example of an external factor, one that had the biggest impact on the labour market in recent memory, is the Covid pandemic lockdown. No one could have predicted the massive disruption caused by the spread of this virus. Alameda County’s decision to extend the lockdown, freezing Tesla’s manufacturing plant, was a near extinction event for Tesla. Elon Musk went into survival mode: he sued the county, threatened to move operations outside California, issued an invitation to all workers to come back to work at the plant on a voluntary basis (if they decided not to, they would still have been paid), and if the authorities wanted to arrest anyone, begin by arresting him first. In the end he prevailed, wining a reprieve, but the scare must have made a deep imprint in his memory. 

Another external factor, which is not that obvious currently, although it is gaining momentum and starting to make the headlines, is the shifting in the attitude of the working population in the US. One unexpected outcome of the Covid19 pandemic is the surge in demand for workers that supply is unable to match. The blue collar workers benefited most from this imbalance: “It’s a workers’ labour market right now and increasingly so for blue-collar workers”. “We have plenty of demand and not enough workers”. Employees are considering changing jobs for better opportunities, more job satisfaction or higher wages (or all of the above).  These jobs are difficult to fill even when the minimum wage is lifted. Many companies offer bonuses on signing up on the job. 

The lack of available workers has had an impact on shipping operations. The ports of Los Angeles and Long Beach, California, that handle a quarter of total US imports are affected by the difficulty to hire enough workers to operate 24/7. 

California has recently passed a law that imposes tougher penalties for companies that put pressure on workers to improve their productivity in large warehouses, such as Amazon’s.  This has a direct impact on the cost of labour, and tilts the balance even more in favour of blue collar workers, which in turn makes it more difficult for companies that lack pricing power, especially in cutthroat competitive industries. A notable trend in all this (another kind of external factor) is the increased demand for higher education. Higher pay is not necessarily enough. Is the work better, does the job offer good prospects of career upgrade, is it safer? These are perks sought by many employees or potential employees seeking to take advantage of this situation.  This is reflected in an unprecedented move by many large companies (such as Amazon, Target, Starbucks) to offer free access to bachelor degree level of education.

The Future of Physical Labour

Sure, it is hard to predict where all this goes over the next few years, but in the context of the developed countries two things are clear: the size of the workforce available for physical jobs is unpredictable, perhaps even shrinking in relative terms, and that we are witnessing a shift toward jobs of higher complexity that require creative and abstract thinking skills, more flexible, more mobile and better paid.  

The inclination to develop policies in the spirit of the Modern Monetary Theory (MMT) that extend social support contributes to a fundamental transformation of the labour structure. Add to this the expectation (belief, or hope may be a better word, but with the same underlying meaning of expectation) that in the not too distant future Universal Basic Income (UBI) will become reality. 

However we may dream about a world free from hard, physical, tedious and repetitive work, we have to accept that this transformational change will not happen unless we achieve an increase in productivity of similar magnitude. 

Funding education to help workers get more complex and better paid jobs is not enough. Assuming, theoretically, that all those people who want to upgrade their knowledge and skills will do so successfully, they will find there are not nearly enough jobs (as many heavily indebted graduates have found out). At the same time physical jobs still need to be done. In order to generate more jobs of higher complexity, companies need higher skilled workforce with proved experience, but that is not available yet. This is a chicken and egg situation: the graduates cannot wait for those jobs to be created, so many of them will take on more menial jobs involving physical labour to make a living, which in turn limits the companies’ ability to grow in new and innovative areas as there are fewer candidates with proven experience that can respond to their call for hire. This is a productivity bottleneck that is hard to overcome as long as a large portion of the total human resources is allocated to physical, low complexity jobs. 

In order to execute the transition from low level physical work to high level complex work these businesses need to undergo a productivity revolution coupled with an upgrade of their business model in the way it employs superior skills and stimulates higher levels of creativity. This is both a technological and cultural change that is very difficult to implement. 

I want to make it clear that physical work is not by default of low complexity. It could either complex or simple, and it could be such that mastering the skill needed to do this work to perfection is a challenge worth taking. If you are a sculptor, a dancer, an explorer, or an athlete you will need to spend many hours to achieve a high level of mastery of the skill of your choice. Complex or simple, performing certain physical jobs still requires a rare combination of high level mental attributes and talent. They will be the exception, rather than the rule in the future economy that we are discussing.

Something else needs to happen in the real economic world, something that takes on the unwanted tasks without increasing costs. In the past that happened by opening up new markets where abundant new cheap labour was available and ready to join the global workforce. The deflationary effect of the additional workforce helped the original countries with world leading institutions and innovative enterprises to create advanced technologies and move their own workforce into higher skilled jobs. The deflationary centres of cheap labour moved over time from Europe and US to South America, Mexico, Eastern Europe, China, South East Asia, and to some extent in Africa. In the same time, immigration in the Western developed countries helped providing cheap local labour.  Lately, this is also more and more difficult as the manufacturing countries of the world now have, rightfully so, their own aspirations of higher standards of living and wanting to create more advanced industries. These countries want their population to have access to better education, work better jobs, more sophisticated and better paid. 

The End of Cheap Labour

We are at the end of the cheap labour runway. If we want to solve the productivity bottleneck we need to get creative and think more broadly. I believe that an important part of a multifaceted solution is the design of a new generation of truly intelligent humanoid robots. This idea goes against the prevailing current thinking according to which robots are a threat for jobs. It is not, if new and better jobs are concurrently being created. Solving the hard labour problem is one of the most important tasks in the near future for all of us, people of this planet,. 

We are on the cusp of a revolution similar to the industrial revolution that saw immense gains in productivity while concurrently the workforce underwent a dramatic structural change.  The sharp increase of the number of workers earning a living wage in the industrial towns and cities was made possible by the long wave of people migrating from rural to urban areas. 

Without the steam engine this would have not been possible. The timing of the technological advance coincided with a compatible social transformation: rural population, especially the young people of late generations who had no access to land ownership (as land could not be divided any further) and who wanted to start a new, more prosperous life that looked possible, so promising, far away from their family rural lands in distant cities, or even countries and continents.  

Peoples’ desire to change the way they work may be unrealistic, as any subjective dream, but it is nevertheless powerful, perhaps even so powerful to cause the change itself.

The same confluence of factors are at work again: younger generations want a better life, better work, better paid doing something more in tune with their aspirations. Few young people, if any, dream of a safe, life time job working on the factory floor. They want freedom, flexible environment, social mobility, opportunities to create and earn enough to live a better life. Peoples’ desire to change the way they work may be unrealistic, as any subjective dream, but it is nevertheless powerful, perhaps even so powerful to cause the change itself. Before the workforce structural change becomes the norm, there is a period of transition when there is a vacuum in the supply of labour that can satisfy the economic demand for it. The vacuum can be tolerable in a long, not so obvious, superficial transition (the number of people making the switch is small and scattered over time), but if the vacuum is substantial and many people start acting on their plan in a short time it can cause severe labour shortages leading to an economic crisis. 

The emerging of expectation of change in the way we work is caused and inspired by a number of factors, among which we can include accessible online education, the explosion of content, social media which has made the world connected, breaking down the geographical boundaries, all of which acts as a powerful catalyst in the adoption of innovation, propagating a different mindset, an upgraded expectation of a lived life. This is global awakening.  

The realisation that this change is possible was accelerated by the Covid pandemic. Before Covid many people wished they could work from home, businesses considered this arrangement, but the change didn’t really occurred at scale until the pandemic. Now, it is nigh to impossible to have a business plan that doesn’t incorporate the option of working from home.  This change in the labour market is one of most disruptive changes in modern history, perhaps at the same level as the introduction of mechanised textile factories early in the industrial revolution, the spread of mass education, or the adoption of automobile. 

As much as the technology has advanced in the last couple of decades it still lacks the big catalysis that supports the transition to a  modern workforce structure.  The transition needs the equivalent steam engine for our times. We have made a great deal of advances in AI, industrial robotics and automation, but the impact is limited because many of the simpler, repetitive tasks require a great deal of movement complexity and educated judgement that are hard to be carried out by machines. Even if they could, the equipment is expensive and it is usually linked to the task in a rigid manner; a small change needs expensive adjustments of the machinery. 

Tesla Bot to Revolutionise Labour

A few years ago, Elon Musk stated confidently in an interview that he sees no reason why the robotised assembly lines couldn’t work at a speed equivalent to a walking man or even jogging. That was in the early days of Tesla 3 manufacturing. However, the reality soon hit back hard. The problems were so frequent, the changes in programming and tooling required to fix requiring so many tests and shutdowns, that Tesla had to abandon the introduction of this approach and get back to the traditional formula using instead skilled workers. These workers are able to learn and adapt to adjustments on the go as the manufacturing process was being “debugged” with minimum disruption. They are also mobile, they can work looking up at a height above their head with a great deal of flexibility and dexterity, difficult to replicate by a ground based robot. However, working long hours in his position leads to bad neck and back injuries. A better solution is needed.

Ground Based Robots at Tesla, Fremont California (Source: Tesla)

Tesla Bot may be the catalyst that leads to new solutions. The evolutionary jump is so large, the technology could be called Robot 4.0. This time, instead of relying on third parties to provide smart components for its automation plans, Tesla has developed its own AI chip and operating system which arguably is the most advanced in the world. The architecture of the new computing system has been at the heart of Tesla vehicles for real time fully autonomous driving with an unparalleled ability to machine learn in complex settings and create its own labels (the same way we do). Now the same system will be at the heart of Tesla Bots. These intelligent machines will be the first humanoid robots capable of understanding real time commands such as  “Please pick up that bolt and attach it to the car with that wrench” (Elon Musk, Tesla AI Day 2021). This is what an young apprentice does.

I submit that Tesla Bot could just be the trigger that marks the beginning of a new era. Yes, it is true we don’t know how Tesla Bot will actually look like and what will it be capable of. This analysis makes a few assumptions based on observations on advances in computing architecture, advanced materials, haptic technologies, and micromechanics. I considered the emergence of neuromorphic hardware that implement models of machine learning mimicking how real neurons work, with superior performance, density and low consumption of energy, and last but not least the extraordinary ability of Tesla to deliver brilliant, world changing disruptive innovations. 

One of the biggest changes in computing architecture is the departure from a model that largely mirrors the just-in-time delivery industrial model. Until recently, microprocessors haven’t changed dramatically from the first model which is based on a CPU sitting in the middle and communicating with peripheral components through a standardised data bus on which any manufacturer could hook into. Newer computing architecture, such as Tesla’s, Apple (M1), BrainChip (Akida), Google (Tensor) are perform machine learning with high speed memory access using the human brain neuronal structure as a blueprint. 

To give a sense how fast this is going and how powerful these systems become, consider the achievement of Google’s AlphaGo and AlphaGo Zero. The first algorithm, AlphaGo was able to beat Lee Sedol, the Go world champion in 2016. In 2017 AlphaGo’s team published an article about the study related to the design of a new algorithm, AlphaGo Zero. This algorithm was capable of beating the chess champion program in four hours after three days of training and self-play, and AlphaGo Lee in eight hours. That was five years ago. Today, Tesla’s AI Dojo supercomputer is able to handle problems that are vastly more complex, the real world of car driving, with a massive of new data collected every second. Its intelligence must be far superior to AlphaGo Zero’s, but notably inferior when compared with its next iteration (currently underway) which will outperform the current version by 10x.  The assumptions that I am making in this analytical essay are not that far-fetched, if at all. I may well be underestimating what comes next. 

If Tesla will deliver robots with the promised capabilities, it will have a huge impact not only on productivity and hard labour, but it will trigger a chain of reactions that will impact our civilization at large.  The revolution will occur both here on Earth and far, far away, on Mars. 

Tesla: Robotised Labour on Earth

Tesla has vertical manufacturing plants in US, China and now in Germany. The goals are ambitious, far beyond “normal”. From 75k units delivered in 2016 to 500k in 2020, Tesla is on track to deliver close to 900k units in 2021, but its aim is much higher, close to 20m in less than a decade. 

Tesla Deliveries from 2016 to 2021(Source: Tesla)

This goal is not just a number. There is ultimate purpose in Elon Musk’s plans, and that purpose is to build a self-sustained settlement on Mars as an expansion of the human civilization. The success of Tesla is vital for two reason: to advance practical technological innovation that has the potential to be used on Mars and to be profitable to generate funds that Elon Musk can invest into the Mars project. Being the richest person on the planet is meaningless to Elon Musk; what matters is that he has the means to fund his dream project without any constraints. Tesla Bots play a dual role: as cheap labour and insurance against the looming labour market crunch and increased productivity, which translates into higher profit margin. From a competitive stand point, the Tesla Bot, if successful, it will establish a moat around Tesla’s auto manufacturing business that is virtually impossible to cross, at least for a decade or two. Elon Musk stated several times that people don’t realise that Tesla’s competitive advantage is its manufacturing which is difficult to copy. Competitors can catch up and design cars that look appealing and make good batteries, but they will find it is much harder to make thinking cars, change the auto industry and its surrounding infrastructure, and do all that at a high net profit margin. What he is not saying is that a highly skilled workforce design intelligent manufacturing plants that use an army of robots to make robots and cars, all based on the same AI platform is what makes Tesla a formidable competitor.  

Tesla Bot Computing Attributes (Source: Tesla)

The manufacturing plants themselves are large computing, moving machines, all part of a breathtaking machine learning industrial eco-system. We should note that the original data feed is collected through Tesla autos that generate new input 24×7. This data stream will be joined by Tesla Bot’s own data stream generating a different type of data, but all contributing to the evolution of Tesla’s computing platform that keeps advancing through fast iterations. Remember, Tesla Bot is built using AI for general purpose robotics, and that includes manufacturing. 

Amazon uses robots that look like moving blocks with rigid arms that move objects around within pre-defined spaces. They are very efficient, but have limited application. In fact they are made only for Amazon’s warehouses.  Boston Dynamics makes more sophisticated commercial robots, Spot and Stretch, which have a higher degree of autonomy, but they have a finite set of programable capabilities. Stretch looks like a more flexible of Amazon warehouse robots, with application limited to moving boxes from containers onto trucks or conveyor belts or vice-versa.  According to the initial specifications, Tesla Bots should be at least one order of magnitude more intelligent. They could work in a human environment and execute complex orders with limited training, adjusting to a broad range of applications. They can literally be classified as machine labour, and at some point in the future required to pay tax.

Auto manufacturing plants are complex. No matter how well designed, the manufacturing process undergoes many adjustments before it reaches its peak performance potential. For example the Grünheide, Brandenburg, manufacturing plant in Germany will start producing cars this year, but it will achieve peak performance in the second half of 2022. Tesla is also anticipating potential labour shortages and is inviting candidates from all over Europe to apply for a job at Tesla, Grünheide giga-factory. 

The manufacturing process can be adjusted much faster with humanoid robots, as they can learn quickly how to execute relatively simple, repetitive, and potentially difficult assignments.  Meanwhile human operators can focus on higher thinking order tasks with big impact on car quality, performance and functionality. This solves the challenging problem of hiring labour in variable and disruptive work cycles. If Tesla manages to successfully bring Tesla Bot into production, it will gain  a huge competitive advantage which will persist for quite some time. 

The Tesla Bot specifications are scarce, but they give us a general picture of how that machine will look like, its size and strength. I suspect the presentation is sending us on the wrong scent to make us believe the machine is generic and open for public use (as suggested, the robot could be ask to do shopping at the grocery store).  

Elon Musk said that a normal person can overpower and outrun the robot. I find this explanation incongruent with his usual precise reasoning. What is a “normal” person?  What about a child? Could a five year old child overpower the robot? Could a strong man overpower four robots that act under the impression, calculated or given as instruction, that that man is a part that needs to be placed in a crucible? Could an incapacitated person run away from a robot activated b

These are obvious questions that come immediately to mind, but in a commercial setting, there are also other thousands more granular questions that need to be answered before a robot like this can be release for public use.  

Why Tesla Bot Is Not Ready for Public Use (Yet)

Elon Musk prefers to stay out of politics. When he was asked if he ever considered the field of genetics he answered yes, but no thanks, because the field is so sensitive to political interpretation. Recently, when asked about his opinion on Texas governor’s mention of him as someone backing the Texan social policies he gave a diplomatic answer and backed away from any further question with a  “I prefer to stay out of politics”.

Elon Musk political avoidance is anchored in a strong sense of pragmatism. His unparalleled ability to pursue goals that either disrupt existing industries or creates new ones would be severely handicapped in a public arena that is influenced by stakeholders with strong political opinions and no skin in the game. He moves at a speed of one thousand miles per hour and wants things done now; he does not want unnecessary, low value high risk distractions that slow him down. 

The idea of a robot available to the public at large doesn’t make sense. Surely he sees what is coming, and what’s coming is going to take up his time infinitely more than the issue of public safety posed by the Tesla’s self-driving system.

Tesla is the undisputed leader in the EV industry. The auto industry regulators are notoriously conservative, so to have cars equipped with the autopilot feature is a big achievement in itself for Tesla. The benefits of this innovation are beyond doubt as statistics show it:

“In the 1st quarter, we registered one accident for every 4.19 million miles driven in which drivers had Autopilot engaged. For those driving without Autopilot but with our active safety features, we registered one accident for every 2.05 million miles driven. For those driving without Autopilot and without our active safety features, we registered one accident for every 978 thousand miles driven. By comparison, NHTSA’s most recent data shows that in the United States there is an automobile crash every 484,000 miles”. (Tesla, Accident Statistics 2021 Q1)

Why does a single Tesla accident make the headlines, while all the other car accidents are ignored? It is not that accidents like these should be ignored, but the asymmetric scrutiny is so obvious. Is it bias? Is it because we resist new paradigms and tend to attribute greater risk to a situation because we are unfamiliar with it? Whatever the reasons are, the fact is sections of society and institutions have a negative perception that often results in investigations, reticence of acceptance, and at times even outright rejection. 

The more social a disruption is, the more open to ad-hoc and uninformed opinions is. Today this is amplified by social media several times over. For good or bad reasons, innovations get different levels of scrutiny depending on impact and perception. In addition to normal inertia, some innovations stir up political games. Disruptors need to invest in the department of public relations, ethics, and law if their innovation is susceptible to social debate and political posturing. 

When decisions are made on consensus, and when consensus is based on status and power incentivised to protect the status quo, change is slow and difficult

For someone like Elon Musk inefficient debates are a major distraction, energy sapping, wasteful activities. It is not because he doesn’t like to engage in broader debates, but the gap between where he is and where others are is so wide, a debate risks becoming unproductive quickly. Persuasion would be difficult, and trying to reach consensus could be very counterproductive. This is one of the reasons why established incumbents are slow in recognising powerful disruptive trends. It took many years before the auto industry finally started to see that Tesla is a real auto company eating their lunch. When Dieter Zetsche announced his retirement in 2019 at the AGM he was asked by a journalist if he fell asleep at the wheel while the EV revolution happened. That was not a question, but rather a statement of fact and frustration. How could have he missed the ginormous transformation? I am sure many people debated behind the closed doors for years when was the right moment to truly start building electric cars at Daimler Benz, but no action was taken. When decisions are made on consensus, and when consensus is based on status and power incentivised to protect the status quo, change is slow and difficult. 

In the early Tesla years, Elon Musk was admired and ridiculed in the same time. He was called a genius and delusional. His quarterly financial results press conferences were labelled “psychedelic” by financial markets commentators. Bob Lutz made a profession out of predicting Tesla’s demise in so many interviews, jokingly referring to a future movie called “Who Killed Tesla?” In the end all these opinions did not matter. Cars are physical exemplars that once you start delivering quality products no one can deny your claim to success. Ideas, concepts, moral judgments on the other hand are prone to stir up different kind of debates, the ones that are won on perception, cultural and political beliefs. 

Self-driving concept is stepping into that realm of subjectivity and it proves to be the target of a difficult debate. Tesla has done a remarkable job at fitting its cars with comprehensive data collection capabilities streaming data that can be used for analysis, improvement of algorithms and for detailed forensic examination of accidents that helps provide objective diagnoses. The science and engineering behind this solution is so far ahead of regulators, but that didn’t stop them from making unfavourable judgments toward Tesla. 

Humanoid robotics is a completely different class of technology posing more complex challenges than ever. Welcome to the universe of unrestricted opinions, perceptions and influence wars. Facebook controversies are child’s play compared with what could happen when humanoid robots are roaming freely in your neighbourhood and your grocery store. Is our society is ready for this now? There is no legal framework ready to deal with incidents, accusations and complaints caused by these robots. This may sound a bit far-fetched, but humanoid robots also need their own rights to protect them from being abused or set-up in doing the wrong thing. Then there is the issue of who has cost, accessibility, and equity. There is a long list of issues, too long to mention here, and I am sure Elon Musk, pragmatic as he is, doesn’t want to have any of these. I submit that Tesla Bots will be limited to use on Tesla’s factory floors.  The whole Tesla manufacturing universe will become an ecosystem of humans and humanoid robots like we have never seen before. This will take the meaning of “Tesla’s competitive advantage is manufacturing” that Elon Musk stated lately on several occasions to a new level. In the EV world Tesla is the incumbent. 

SpaceX: Robotised Labour on Mars

This is the most compelling raison d’être for Tesla Bots. Mars is their true calling. 

Imagine you are the first human to step on Mars. You land, crack the door open and you see this: 

Mars June 2021 (Source: NASA, Perseverance Rover)

Where do you begin? Unpacking the suitcase? There is a myriad of things to do just to set foot on the planet, so to speak. Dig a ditch to protect your crew from radiation, set up a tent, organise the supplies, install appliances. So much to do. It is huge. It is hard work.

We Choose To Fly to Mars not Because It Is Hard, but Because It Is Necessary

The goal of the first trip to Mars is so unlike the first trip to the Moon. In 1969 the mission was simple: survive, pick up some rock, take a few steps, get back to the station and fly back to Earth.  The project achieved a psychological, uplifting goal in a competitive global race for influence. JFK’s famous words “We choose to fly to the moon not because it is easy, but because it is hard” doesn’t apply today. An updated version of this quote adapted to our time would say “we choose to do hard things because they are necessary”. The Mars project is vital for the survival of our humanity in the long term. The first trip to Mars has a massive mission: establish the first outpost on Mars. Survive and prepare. From this moment on the operation “Populate Mars” begins.

Astronauts on Mars Should Grow Settlements not Plants in a Greenhouse

The striking aspect of most of the artistic representations of future life on Mars is the absence of robots. Life looks comfortable, astronauts are enjoying the view as if they are tourists.

Source: Tesla

Most of the commentary regarding the future settlements on Mars are concerned with food production and psychological effects of isolation. The many experiments replicating life on small scale bunkers built at remote locations are focused on these two aspects: food production and mental health.  All these concerns are legitimate, but they are oblivious to the need of infrastructure required to help the first settlers deal with these concerns. Robots are out of the picture.

I recently re-read Isaac Asimov’s Foundation. This has been one of my favourite books. I love the idea of psycho-history, but I am always deflated by the oversimplistic background of the story. here are a few huge buildings with bad people attempting to control everything. There is no robot there. The Three Body Problem has a few intriguing concepts, but again no robots, just people and a huge controlling software. Even the Mars trilogy (Red Mars, Green Mars, and Blue Mars), one of the most detailed and well researched Sci-Fi’s, makes little mention of robots. It occurred to me that we are still thinking under the influence of the IBM mainframe style of thinking. Odyssey 2001, with its mysterious metallic block influencing the evolution, has at is core HAL, and IBM creation.  This is not what will happen in reality on Mars, and I am explaining below why.

There Is a Lot of Work to Do

The first Starship will transport about 100 people to Mars. This is an incredible difficult project. They have to start building things straightaway (this scenario applies to the whichever mission that begins the settlement process).  If you only consider the mundane actions taken through the day just to maintain a safe and stable living environment, without help there is little or no time that can be dedicated to building larger structures that are vital to the establishment of the first settlement. Definitely there is no time for research, development and design projects that can only be done on Mars and for the people on Mars (Martians!). 

The ship is designed to transport about 100 tones of stuff. The first priority is to build structures that produce energy and methane necessary to fuel the trip back to Earth. Then you have to build oxygen production plants, some sort of structures which could serve as storage and living quarters. This is far from a complete description of  the stages of the initial “Populate Mars” project , but I am trying to give an idea of the gargantuan work these 100 people will be required to do; it is overwhelming.

Ars Technica describes potential deliveries carried by the first two cargos to Mars: massive array of solar panels, mining equipment, surface vehicle, food and life support infrastructure, a propellant production plant and a greenhouse assembly parts, solar park and landing pads. Who will move around 200 tons, assemble the plants, and grow food? It is hard to believe the first 100 people would be able to do all that. 

The Organisational Structure on Mars Enterprises

Initially, it will be one group of people who need to do stuff. That is a small enterprise, but enterprise it is because it needs to build things, produce planned outcomes, expand and, more importantly, solve problems that require high level of thinking and advanced knowledge. This is an applied innovation enterprise at max. 

Innovation Workforce Infrastructure

A typical innovation organisation on Earth has a structure predominantly made of highly skilled and creative individuals. That is what you see on paper. In reality, each of such enterprise is supported by a large workforce that is scattered across other various organisations specialised in providing services developed long time ago in which they excel in execution speed and efficiency. This is a workforce infrastructure that is not visible on the organisation chart of innovative enterprises because it is an external resource. Road maintenance, transportation, healthcare, engineering and construction services, energy, electricity, logistics, communication etc. they are implicitly part of an innovation enterprise, although for practical purposes they don’t figure in any activity report, except as a number in the expense column. 

On Mars this infrastructure is non-existent, which makes it nearly impossible to innovate. You could send more humans, but that just increases the scale of the problem, not making any substantial progress. This is why humanoid robots are a vital addition to the human team: they can play the role of the workforce infrastructure we are taking for granted on Earth, at least partially. 

Artificial Workforce Infrastructure

Of course, this artificial workforce infrastructure will not be ready on the first day on Mars. The first generation of humanoid robots will be needy. Tesla Bots are not (yet – ha, ha, ha) cyborgs, capable of making independent decisions with human like behaviour, so they will need to be “installed”, instructed, supervised, maintained and repaired. Depending on how capable these robots are, the first human teams on Mars must have a small group of robotics engineers who know Tesla Bots down to the last bit.  Their job is to immediately unpack them, and set them to do the first tasks: carry boxes, dig, assemble parts, build energy and oxygen production units, build repair stations, set up tents or small buildings, etc. Before flying to Mars the robots must have some training done on Earth to have learned a few fundamental skills (such as pick up a wrench, a hammer, a ground digging tool, a drill, etc.) Maybe a tenth of the initial group should have these engineering skills, and they should be able to control and manage a team of 100 robots, or maybe more. 

The human civilization on Mars will be the most robotised society in human history

Each robot weighs around 75kg, so in total, a manageable humanoid robotic team would be 750Kgs, about the same load as the human team. The mission could easily add another 200 robots packed in compact containers as they don’t need oxygen and space to move around to stay fit. We could assume that the breakdown rate could be high initially, so a sufficiently large reserve of robots is necessary to help maintain a functional humanoid robotic team until the mini-colony is set and stable. If some of these robots could be trained to do a high level supervision of other robots, the ratio of active robots to humans could be even higher. One of the key features that Elon Musk mentioned in his presentation that a human (adult) can easily overpower a robot is especially relevant to Mars. In case the robots get the wrong idea there is no 911 to dial there. This aspect of security and safety will be a permanent concern for the Mars settlement. The human civilization on Mars will be the most robotised society in human history. Asimov’s three laws of robotics will be written in the first Martian constitution to guarantee protection of humans not only from malfunctioning robots, but also from robots programmed by malfunctioning humans. 

With each mission a group of robotics engineers is sent to Mars with new knowledge and a new generation of robots. Some of these engineers will have the knowledge and skills to  building robotised factories that make robots increasing dramatically the productivity in the Martian settlements. The key point here is machines have to do repetitive work at scale so that highly qualified people can focus on solving complex problems once. In an interview to Ars Technica (in 2020) Elon Musk makes the distinction between what engineers can and should do and what machines can and should do

“If you’re just trying to make one of something, it can all basically just be made by the engineering team,” he said. “But if you want to actually make something at reasonable volume, you have to build the machine that makes the machine, which mathematically is going to be vastly more complicated than the machine itself. The thing that makes the machine is not going to be simpler than the machine. It’s going to be much more complicated by a lot. Things need to be translated into instructions that the average person can understand. You can’t have somebody with an engineering master’s degree from MIT hand-making every single part. It’s not possible. There just aren’t enough. MIT’s not graduating enough people.”

This will be a fascinating chapter of human history with massive changes on both planets. 

If Musk is right, one thousand ships will fly to Mars in large fleets. Mars will be populated with ten thousand people in the space of  one or two decades. This prediction is based on little evidence, other than the extraordinary work at SpaceX that would have been considered Sci-Fi ten years go. Well, it even looks sci-fi as we speak, but we haven’t had that feeling yet because it is difficult to comprehend the scale of the operation and visualise these rockets flying to Mars until you actually see it happening. 

Who Will Fly to Mars?

There is something else: we have the tendency to say “we SEND people to Mars”. It sounds like we catch unaware pedestrians, put them in a box and send them. That is not how it works, not that it hasn’t done before, but this is not Australia in 18th century. People who fly on a three months journey have to want to go there, but why would they? Despite knowing that life on Mars will be hard, there will be many volunteers who would take this as an amazing opportunity to be the first people to land on Mars, establish a new colony and break new ground. They will make history. But, like the fascination with which the world followed the US Mars unmanned vehicles land,  looked at the first photos, and watched the first helicopter fly, the interest will dwindle after a while. 

Who will go up there then? I believe that will be the people who want a new career, people who want to experience a new life, to do something extraordinary by doing what they know best. That is not going to happen if most videos sent from Mars show people working nearly full time growing carrots and tomatoes inside non-descript greenhouses, or walking in a red desert struggling to make their way through air filled with red sand spurned by random cold winds just so they can check the communication beacon is OK. Look at what happened to Matt Damon; it can happen to you.

If we are to expand the settlement on Mars “normal” people have to feel they wish they can go there. Half of the population will need to be made up of women who want to consider living there for a long time, have family and children. Civilization on Mars must build a culture based on a diverse pool of engineering, scientific and creative talent capable of writing their own artistic stories, make their own Netflix, fun parks, tourism and discoveries. We need to establish early on a cultural settlement, not just technological, not just building. If it is all work and no play, who would want to go there? 

Robots. Labour. We need robots, lots of them, humanoid, smart, versatile, capable of completing hard tasks with minimum supervision. They are essential for the Martian settlement to survive, grow and flourish into a brand new civilization. Although this civilization is designed on Earth it will become a very different reality on Mars. It will be free from any reticence toward the use of humanoid robots. From start, the human population will be much smaller compared to the population of robots. One day the settlement will reach a level of technological and manufacturing capability when they will design and make their own robots. Humans and humanoid robots will have a symbiotic relationship. If this pocket of civilization survives and reaches that level it will grow at a speed and in a form we cannot even imagine today.  That will be the day when Tesla Bot will look as archaic as the first transistor. 

This analytical essay revolves mostly around Tesla Bot, Tesla, SpaceX and Mars. However, I am convinced that as soon as the humanoid robot is taking shape, the competition (I cannot imagine Apple not being in this business) will spring into action. This time, they are not going to wait and see as they did in the early days of Tesla. The first wave of humanoid robots will be used in manufacturing and other heavy industrial activities. The second wave will make its way into the public domain with applications in healthcare, retail, household, and hospitality. Once this revolution starts it will move really quick.