Photo by @alesnesetril

Artificial Intelligence and Indentured Labour

FourthBrain Machine Learning Bootcamps

Would you commit to an unknown company for five years in exchange for becoming a machine learning junior? I will be exploring the case of FourthBrain that announced an employment scheme in September 2019.

It sounds so archaic, old for some, not really part of the present. Then again indentured labour is a practice that never really went away, much like saying imperialism is a concept of the past — of course it is not. Are technology companies so blatantly ahistorical that they do not understand their framework is not novel? The answer seems yes. At least as an anthropologist I can argue there are some similarities worth exploring, which I will do here. I may be operating in a grey area, and many would argue that I am wrong (as I may be) however I strongly encourage discussions on this topic. The definition of indentured labour can be said to be the following:

An indentured laborer is an employee (indenturee) within a system of unfree labour who is bound by a signed or forced contract (indenture) to work for a particular employer for a fixed time. The contract often lets the employer sell the labor of an indenturee to a third party. Indenturees usually enter into an indenture for a specific payment or other benefit, or to meet a legal obligation, such as debt bondage.”

“Just as electricity transformed numerous industries starting 100 years ago, AI is now poised to do the same. AI Fund is a startup studio building new AI companies from the ground up.”

The commitment is a full-time, six-month residential training program with the goal of training software developers to be ‘junior machine learning engineers’. It starts in the fall 2019. There are no upfront costs. In addition, they will provide you with housing and onsite meals, for the entirety of the program.

They are looking for students with a hunger to learn and ability to focus full time on studying machine learning. The applicants do not need to have any prior background or experience in machine learning. What you do need however is still considerable:

“Engineers and Data Scientists who have had at least 2 years experience in their respective fields and a basic grounding in coding from on the job experience. The most successful candidates will have experience with the following: — Demonstrated strong coding from previous work experience or publications. This means you’re able to write a non-trivial program in Python, Java, or C++. — Solid CS foundation (including but not limited to Operating Systems, Computer Networks, Database, etc.)”

The program is paid by ‘future employers’ and you are expected to work hard throughout the program as is shown in the application form.

The job advertisement says nothing of the length which you commit to any future employers, however through a group on Facebook this was mentioned clearly. As an ‘income share agreement’ over the course of five years is to be committed.

You will pay back the program cost through an income share agreement — a portion of your future income over 5 years. This has to be written more clearly during the course of the application process. However it is not and it is of course a cause for worry.

What is said to be different from indentured labour is bonded labour. The ‘Bonded Labour Abolition Act’ very clearly abolishes the practice of debt bondage — where men or women pledge their labour as a collateral in a debt agreement. Bonded labour is illegal in the US. This however may not as clearly be the case elsewhere, and practices of what constitutes bonded labour may have a few grey areas. FourthBrain can therefore be argued to be indentured labour as it can be said to be freely committed as you sign up to a program, not as part of collateral debt obligations.

It may be unfair to frame it even as indentured labour, perhaps it can be seen as an apprenticeship, given the benefit of doubt? An apprenticeship is a system of training a new generation of practitioners of a trade or profession with on-the-job training and often some accompanying study. I can attempt to categorise these as the following:

Clear and unclear view of what an apprenticeship is.

Therefore we could argue that the limit in the proposal by FourthBrain is clearly defined, and therefore it is not bonded labour. However there is a fine line between apprenticeship and bonded labour or slavery that seems to have been crossed in the past.

Banner against apprenticeship from 1834 retrieved from Discovering Bristol on Tuesday the 1st of October

More recently there has been discussion too of the limits of apprenticeship and perceptions in regards to this whether it is an opportunity or exploitation. Apprenticeship is a kind of job training that involves following and studying a master of the trade on the job instead of in school.

In the end of September 2019 a job ad was posted on LinkedIn looking for a CEO to run FourthBrain. This has thus been put out while the search for candidates for the ‘machine learning bootcamp’ was announced. This is a full-time position based in Palo Alto, California. Within the description of the recruitment ad it is stated: “You’ll build and manage a world class team to scale all aspects of the company, in order to serve thousands of students.”

As such it is arguable whether the program provided with its scale is such an organisation to provide the personal follow-up or facilitation expected from an apprenticeship scheme. Independent apprenticeship in the previous categorisation showed in the figure still means follow-up it simply means independent from any educational institution. This can however be loosely defined, and perhaps it has changed today.

Tata consulting and Cisco Systems has claimed that contracts signed in a different country has a different validity, and there are already cases of bonded labour in the United States within IT. It is attractive to move to another country, learn new skills and have the promise of a job afterwards. In fact it sounds like a great proposition.

So, exactly how much do machine learning engineers make? The average machine learning salary, according to Indeed’s research, is approximately $146,085 (an astounding 344% increase since 2015).

The average machine learning engineer salary far outpaced other technology jobs on the list. A writer in Springboard displays the stats shown by LinkedIn’s U.S. Emerging Jobs Report, the demand for MLEs increased by nearly 10 times since 2012.

LinkedIn’s U.S. Emerging Jobs 2017

Working in machine learning is a much coveted job opportunity. If you are a software engineer already working in some shape or form with programming machine learning and artificial intelligence. As the demand and growth is so high it can have led to an increased focus on bringing in new talent quick, as well as seeing this as a business opportunity in itself.

What they are looking for in FourthBrain is someone who has experience as a data scientist, yet wants to transition into the role/title of a machine learning engineer. On that note the AI Fund startup studio recently expanded to Colombia. They do have an expressed intention to train ‘millions of AI engineers’.

“From Silicon Valley to Puerto Rico, students now have the opportunity to learn next-generation artificial intelligence applications, theories, methods, and systems. Our team, mostly from Stanford, has collectively educated more people in machine learning than any other organization in the world. The program is a blended-learning program that trains existing engineers to become machine-learning experts within six months.”

The AIFund was set up partly by Andrew Ng who is a famous machine learning expert who additionally set up one of the most popular online learning platforms in the world called Coursera. He is a Chinese-American computer scientist and statistician, focusing on machine learning and AI. Ng is an adjunct professor at Stanford University (formerly associate professor and Director of its AI Lab). Since 2018 he launched and currently heads AI Fund, initially a $175-million investment fund for backing artificial intelligence startups. On the surface FourthBrain seems to be one of these startups.

The goal of FourthBrain stated on their website is to: “build Puerto Rico into a world-class AI Hub.” Further they explain: “To help curb global inequality by training and placing members of underrepresented populations in highly compensated machine-learning jobs. We believe in brain circulation, not brain drain.”

On the website there is a clear outside-in perspective appealing to the Puerto Rican.

“To help transform Puerto Rico into the world’s AI Island. We know that as a Puerto Rican, you are proud of your people’s history, culture, and heritage. You have made enormous contributions to the world, including fighting in numerous wars to make the world safe for freedom and democracy. We also know you have had to overcome, most recently, a quasi-bankruptcy, an economic depression, being battered by one of the most destructive hurricanes, and rampant graft and corruption. You seek respect and economic opportunity. We know because some of us are Puerto Ricans too, and those of us who aren’t feeling a special kinship to a people who have sacrificed so much and whose contributions have so often gone unappreciated. We are confident that by learning AI, you will leapfrog the Island to a better future.”

We (company, Stanford people) <> You (Puerto Rican) according to what I have seen so far seems like an interesting way to display this. Why not just give people a job, payment and training? As a previous employer in a small company I understand it is difficult to bring in new employees and budgets needs to be balanced etc. However this does seem like a strange practice, especially when sold/pitched this way. I should probably not speak to loudly in opposition to this as a privileged, white and Norwegian university student. I would however ask you what do you think this sounds like?

Screenshot from Fourthbrain.com retrieved the 1st of October

Please excuse my gif. reaction, because this sounds too amazing. Wow.

Endless possibilities, wow. I wish I had 2 years experience in Data Science.

In an ironic sense, it seems making an impact holds an importance. It can be argued that those working in the field of artificial intelligence has the chance to contribute, as members of society with a unique skill. On the other hand, ‘endless possibilities’ in a five year work commitment after a free training sounds slightly over the top. In the recent ousting of the WeWork CEO Adam Neumann the marketing professor Scott Galloway termed this type of language ‘yoga babble’ inflating the value of the activity of a given company.

The field of artificial intelligence has been such an interesting field to explore. It seems almost unholy (although I am not religious) to associate or even hint of negatives relating to such a hyped topic. Certain areas or application of artificial intelligence seems upright and some less so, much in the same way AI in relation to the defence industry or security is talked of. It may be strange as well to criticise an initiative to educate more people from different backgrounds — as on the outset this is an important ambition.

However I think we should have no illusions as to the possibility of labour practices or other illicit ways in technology, such as the MIT Media Lab scandal in September 2019 concerning Jeffrey Epsteins donations. Epstein, who was found dead in his jail cell after being arrested on sex trafficking charges, donated almost $10m to MIT over two decades, with many of the gifts coming after he registered as a sex offender in Florida in 2008.

It is said by Anna North in Vox that this placed a spotlight on human trafficking. The federal government in the United States defines trafficking as using “force, fraud, or coercion” to make someone perform labor. That can include sex trafficking. Epstein was facing trafficking charges in connection with allegations that he paid underage girls for sex when he died.

A letter written by both female members of the faculty and former assistants to MIT read:

“How can MIT’s leadership be trusted when it appears that child prostitution and sex trafficking can be ignored in exchange for a financial contribution?”

We should perhaps start exploring and questioning even further practices in technology beyond funding or who gives the money to research projects. We can question the labour practices in technology, and the field of artificial intelligence in particular with this indentured training in machine learning techniques as an example.

I have found it interesting to discussions in other cases to look at historical sources, although we are talking of a different context. I managed to find a few examples of ads for indentured labour in the past perhaps worth taking a look at for comparison.

An advertisement in a Philadelphia newspaper. In the 17th century, nearly two-thirds of English settlers came as indentured servants. You can here see the indentured labour alongside the ads for slavery.

Indentured labour was not only for those who had little working experience, quite contrary, even for clerks indentured labour was a possible way to go. Of course an attractive proposal for those who were well educated and wanted to see the world or gain access to new networks.

An advertisement from the newspaper Glasgow Courant, 4 September 1760, for indentured servants to go to Virginia.

On the other I do believe, somewhat the working conditions of a machine learning junior is different. Yet both academic (above) and technical skills (below) were required in the past as well, so this at least is nothing new.

An advertisement from the Glasgow Journal, 1 September 1763, for a blacksmith to go to the British colony of Maryland on the east coast of America.

An indenture was a legal contract between a master and a servant. The servant had to work for a set period (most commonly seven years) with the master agreeing to pay his passage to the colony, food and accommodation and, occasionally, a small salary.

After I did some searching as previously mentioned I discovered that with FourthBrain you commit to five years. During the six months you get food and accommodation. Later when this is completed you get a fair salary, it seems based on the grace of your future employer.

The length of servitude was not mentioned in the FourthBrain advertisement online, at least the ads from the 17th century had the courtesy in one of the mentioned cases of stating the four years of servitude needed.

The two other ads from the 17th century ad say ‘for further particulars’ or mentions ‘further encouragement’ perhaps this can be compared to FourthBrain. The terms of the contractual relationship is not stated either. In past history these conditions were explicit:

“The servant was not allowed to leave the service of his master until the end of his contract, although the latter could sell the remaining period of the indenture to a third party […] There could be severe penalties for servants who tried to break their contracts.”

In a previous case of IT companies in the US the companies loosing bonded labourers sued their past employees and won. In this context I wonder what the penalties will be like in a stated contract. This would be interesting to better understand as it is not mentioned online or talked of by FourthBrain on their website.

The indentured labour was purchased from a sea captain in the past. However venture capital often has the legal or business finesse to bring labour into different areas of the world, such seems the case too in the case of FourthBrain. The employer purchases the indenture from FourthBrain which brings the talent over to Puerto Rico on Colombia to work then to be bound to a five year contract with an employer. For those with limited means indentured labour can mean an opportunity to travel and establish yourself, once free of indenture, leave with new skills for a better life. We may be sceptical of indentured labour in technology as much as we are sceptical of this practice in other fields or areas.

Side note:
I guess if the alternative is to pay a sum for a YouTube Start to teach you, indentured labour may be a viable option? Raval on his Make Money with Machine Learning infamously overcharged and lied to those who participated. It is a contentious story in the AI community online.

As is expressed eloquently in this meme there are however other free options online to learn the basics of machine learning techniques:

Meme posted by Kev Muriithi in ‎Artificial Intelligence & Deep Learning retrieved 1st of October 2019

There are surely more dubious or at least questionable practices within the machine learning community and more broadly the field of artificial intelligence that can be explored. However we should additionally try to understand the other side of this story and not only my surface judgement written out in this article.

This is day 120 of #500daysofAI. If you enjoy this article please give me a response as I do want to improve my writing or discover new research, companies and projects.

AI Policy and Ethics at www.nora.ai. Student at University of Copenhagen MSc in Social Data Science. All views are my own. twitter.com/AlexMoltzau