The history of the future of automation – lessons from ed tech

I’ve been re-reading the excellent essay by Audrey Watters: Driverless Ed-Tech: The history of the future of automation in educationI really recommend reading it!

Following on from the vague points I made in my last post about an automative imaginary – one way of getting at this is a sort of classic (political economy) critique, which Watters very adroitly takes in relation to ed-tech and MOOCs (etc), and which she write really clearly – so, here is snippet:

“Put me out of a job.” “Put you out of a job.” “Put us all out of work.” We hear that a lot, with varying levels of glee and callousness and concern. “Robots are coming for your job.”

We hear it all the time. To be fair, of course, we have heard it, with varying frequency and urgency, for about 100 years now. “Robots are coming for your job.” And this time – this time – it’s for real.

I want to suggest – and not just because there are flaws with Uber’s autonomous vehicles (and there was just a crash of a test vehicle in Arizona last Friday) – that this is not entirely a technological proclamation. Robots don’t do anything they’re not programmed to do. They don’t have autonomy or agency or aspirations. Robots don’t just roll into the human resources department on their own accord, ready to outperform others. Robots don’t apply for jobs. Robots don’t “come for jobs.” Rather, business owners opt to automate rather than employ people. In other words, this refrain that “robots are coming for your job” is not so much a reflection of some tremendous breakthrough (or potential breakthrough) in automation, let alone artificial intelligence. Rather, it’s a proclamation about profits and politics. It’s a proclamation about labor and capital.

More specific to education, Watters highlights the logic offer by Udacity, one of the big MOOC start-ups:

“We want to be the Uber of education,” Thrun told The Financial Times, which added that, “Mr Thrun knows what he doesn’t want for his company: professors in tenure, which he claims limits the ability to react to market demands.”

In other words, “disrupt” job protections through a cheap, precarious labor force doing piecemeal work until the algorithms are sophisticated enough to perform those tasks. Universities have already taken plenty of steps towards this end, without the help of algorithms or for-profit software providers. But universities are still bound by accreditation (and by tradition). “Anyone can teach” is not a stance on labor and credentialing that many universities are ready to take.

[read all of Watters’ argument here.]

One of the areas in which what I’ve called the automative imaginary/imagination overlaps with wider stories about digital tech disruption and the widespread creed of that disruption being normatively good, if not admirable, is where automation overlaps with the ‘gig economy’. As we can see, via Thrun, one of the models in play here is not to ‘destroy’ jobs but rather to decouple them from some of the elements that make them valuable and either reallocate or automate those and then offer the remaining tasks as precarious work – which you have to sign up for through a proprietary system as a contractor, rather than an employee.

The ways we are invited to see ‘automation’ by some are over-coded with strong narratives of ‘disruption’ and the breaking down of established employment rights into a two their system in which the ‘taskers’ (as Guy Standing calls them) take the precarious work for relatively meagre earnings and the tech company owners and a few they bring along with them hoover up the profits.

An alternative to such narratives might be ‘accelerationism‘ but I remain skeptical at present (but I’ve not read enough to form an opinion really). It seems to me, we (geographers, social scientists, citizens?! … well, me anyway!) really do need to talk and think about work.

More tales of the automative imaginary

Here’s some links that further sketch out some of what I’ve been thinking about as an ‘automative imaginary’. I’ve offered links with a bit of brief commentary at the bottom…

Evidence That Robots Are Winning the Race for American Jobs – in the NYT, pointing to research undertaken by two economists, Acemoglu and Restrepo, published by the (American) National Bureau of Economic ResearchRobots and Jobs: Evidence from US Labor Markets (to which I have no access), with a commentary on the Centre for Economic Policy Research‘s Vox site. What is curious for me here is how one can evaluate the method of the researchers and what the assumptions they make say about how we (are invited to) understand automation. There’s some interesting geography in there too! E.g. see the choropleth map of “exogenous exposure to robots” below

How will the rise of automation and AI affect the workforce and economy moving forward? – Francis Fukuyama offers his answer to how automation and AI (interesting easy slip between those as almost a form of equivalence, which is open to significant debate/critique) may or may not “affect” the economy and, in particular, jobs – in the US.

It’s interesting how much of what we are offered in terms of a rationale for automation is a fairly simplistic robots replace workers sort of story. In this regard, it’s worth remembering what the MacDonalds CEO Ed Rensi flippantly observed as a canonical example (documented in this post on Fusion):

former McDonald’s CEO Ed Rensi made news by going on Fox Business and declaring that ongoing protests in the campaign for a $15 minimum wage were encouraging the automation of fast food jobs. The segment goes on for seven minutes, but here’s the meat of it:

I was at the National Restaurant Show yesterday and if you look at the robotic devices that are coming into the restaurant industry — it’s cheaper to buy a $35,000 robotic arm than it is to hire an employee who’s inefficient making $15 an hour bagging French fries — it’s nonsense and it’s very destructive and it’s inflationary and it’s going to cause a job loss across this country like you’re not going to believe

Nevertheless, other economists will tell you that processes of automation have, historically, created new kinds of jobs as they apparently ‘destroy’ others. For example, Deloitte, in their report “Technology and People: The great job creating machine“, suggest that while manual labour and routine jobs have been significantly automated since 1992, there has been an even larger growth in ‘care’ (and service) and ‘cognitive’ work in the UK labour market. So you see fewer people in manufacturing but more analysts, baristas and carers.

Of course, to see it as whole “jobs” that are being automated is somewhat misleading – another aspect of the automative imaginary that owes more to the depiction of automation in 1950/60s cartoons than in the actually existing forms of automation. As many commentators point out, it’s parts of jobs or tasks that become automated, which results in a need to reorganise that work. As the management consultants McKinsey point out in a report in 2016:

currently demonstrated technologies could automate 45 percent of the activities people are paid to perform and that about 60 percent of all occupations could see 30 percent or more of their constituent activities automated, again with technologies available today

What we tend to focus on is the automation full stop, not that it isn’t all of a job and may not result in an easy equivalence of “robot in = worker out”. We imagine the robots doing it all, when, in many cases, the use of robots (when they’re actually economically viable – they have a huge initial set-up cost) require a reorganisation of systems such that the work looks different.

Another illustration of this comes from the excellent Containers podcast by Alexis Madrigal. In the final episode, Madrigal talks to Karen Levy of Cornell  about the forms automation could take in relation to truck driving (upon which Uber clearly has its sights set). Of course, again, it’s not as simple as: automate the lorries, do-away with jobs. It’s more like the process of containerisation that Madrigal is exploring – automation is as much about reorganising systems of work / labour as it is about ‘replacing’ labour. So, in the example of picking in warehouses – you might get a Kiva or Fetch/Freight robot to do the donkey work of warehousing, with the worker performing the more sophisticated movements. This is not a future of people-less spaces but rather robots following people around or being tasked in order to support the worker, the argument being this leads to greater productivity. In fact, in the eighth episode of Containers, the CEO of Fetch Robotics justifies her company’s tech by saying that, in the US, there are over 600k jobs going unfilled in warehousing and manufacturing because people don’t want to do them, with a turnover rate of those who do sign-up for such work at around 25% (I don’t know the basis or veracity of those numbers – would like to though!). Again, if true, such figures are another aspect of the expectations of what work involves and how it may be performed.

So, it seems to me we need to talk about work not simply elide it by (somewhat hysterically) referring to ‘automation’ and ‘robots’. This is something I hope to research and write more about, if I ever get the time…

Songs “written by AI” from SonyCSL

Songs written by Sony CSL’s “AI”…

From the Sony CSL “flow machines” website:

Flow Machines is a research project funded by the European Research Council (ERC) and coordinated by François Pachet (Sony CSL Paris – UMPC).

The goal of Flow Machines is to research and develop Artificial Intelligence systems able to generate music autonomously or in collaboration with human artists.
We do so by turning music style into a computational object. Musical style can come from individual composers, for example Bach or The Beatles, or a set of different artists, or, of course, the style of the musician who is using the system.

Their “Deep Bach” thing was doing the rounds at the end of last year, so I presume there will be more to come.

A Universe Explodes. A Blockchain book/novel

Thanks to Max Dovey for the tip on this…

This seems interesting as a sort of provocation about what Blockchain says/asks about ownership perhaps, although I’m not overly convinced by the gimmick of changing words such that the readers unravel, or “explode” the book… I wonder whether The Raw Shark Texts  or These Pages Fall Like Ash might be a deeper or maybe I mean more nuanced take on such things… however, I haven’t explored this enough yet and it’s good to see Google doing something like this (I think?!)

Here’s a snip from googler tea uglow’s medium post about this…

It’s a book. On your phone. Well, on the internet. Anyone can read it. It’s 20 pages long. Each page has 128 words, and there are 100 of the ‘books’ that can be ‘owned’ . And no way to see a book that isn’t one of those 100. Each book is unique, with personal dedications, and an accumulation of owners, (not to mention a decreasing number of words) as it is passed on. So it is both a book and an cumulative expression of the erosion of the self and of being rewritten and misunderstood. That is echoed in the narrative: the story is fluid, the transition confusing, the purpose unclear. The book gradually falls apart in more ways than one. It is also kinda geeky.

Picking at the “alternative media ecosystem”

Two things on how various intersecting discourses are coming at “fake news’…  I find it interesting to see and hear how different folks are attempting to make sense of an apparent phenomenon, it’s a little like watching Foucault’s ‘discursive formation’ in action…

First, via Adrian J Ivakhiv:

Parsing the “alternative media ecosystem”

An interesting forthcoming article by University of Washington researcher Kate Starbird examines the “alternative media ecosystem” by focusing on the production of the kinds of narratives that are fairly exclusive to the “alternative,” as opposed to mainstream, “media ecosystem.” Specifically, the piece analyzes conspiratorial narratives, found on Twitter and connected web sites, that follow terrorist incidents (including the 2013 Boston Marathon Bombings and the downing of Malaysia Airlines flight MH17) and several mass shooting events.

“For each event,” Starbird writes, “rumors claimed the event had been perpetrated by someone other than the official suspects—that it was instead either a staged event performed by “crisis actors” or a “false flag” orchestrated by someone else.” (For more, see the Seattle Times and Starbird’s own summaries of the research.)

From Starbird’s scholarly article:

“After several rounds of iterative analysis to identify commonalities and distinctions across clusters of accounts, we identified three prominent political agendas: U.S. Alt Right, U.S. Alt-Left, and International Anti-Globalist.”

Second, via the ‘Team Human’ podcast, hosted by Douglas Rushkoff, Caroline Jack on ‘propaganda’:

What’s Propaganda Got To Do With It?

If “propaganda” is a useful as a media epithet because it expresses concerns about media persuasion and power, then we must allow that a variety of actors, not just states or would-be states, can influence the television networks, newspapers of record, and leading online news sources.

Our understanding of media power (and of what it means to call something propaganda) must make room for a variety of potential collective and individual influences. This includes corporations, interest groups, activist groups, and other traditional collectives; it should also include the new forms of individual and collective presence that digital communications facilitate. This includes state-sponsored online actors and ad-hoc user collectives.

“Control, Resistance, and the ‘Data University’: Towards a Third Wave Critique”

A group of academics at Newcastle, collectivised under the moniker “the Analogue University” offer an Alex Galloway-like critique of “The Data University” over on the Antipode blog. An interesting read…

In this short intervention, we want to explore the possibilities for a third wave of critique related to the changing nature of academia. More specifically, we argue that we are now witnessing the emergence of the “Data University” where the initial emphasis on the primacy of data collection for auditing and measuring academic work has shifted to data coding itself as the new exchange value at work and productive of new subjectivities and freedoms. This third wave critique requires drawing a schematic line that now takes us beyond the intensification of neo-liberalisation, the internalisation of market values and associated affective structures of feeling to understanding our new digital and big data world. Influenced by Deleuze’s (1992) work on new societies of control, we argue that the genesis of the “Data University” lies in our active desire for data and its potential to mediate human relations and modulate our freedoms. This is absolutely central to our schematic for a third wave of critique: compared to older disciplinary societies like the school or prison institution (see below), today individuals both desire and are controlled through the active generation of proliferating data streams.

Read the full article.

Interesting papers: “The Taking Economy” & “The Myth of the Sharing Economy”

These look interesting…

Via Frank Pasquale:

The Taking Economy: Uber, Information, and Power

Ryan Calo
University of Washington – School of Law; Stanford University – Law School; Yale Law School

Alex Rosenblat
Data & Society Research Institute

Sharing economy firms such as Uber and Airbnb facilitate trusted transactions between strangers on digital platforms. This creates economic and other value and raises a set of concerns around racial bias, safety, and fairness to competitors and workers that legal scholarship has begun to address. Missing from the literature, however, is a fundamental critique of the sharing economy grounded in asymmetries of information and power. This Article, coauthored by a law professor and a technology ethnographer who studies the ride-hailing community, furnishes such a critique and indicates a path toward a meaningful response.

Via Tom Slee:

The Myth of the Sharing Economy and Its Implications for Regulating Innovation

Abbey Stemler
Indiana University – Kelley School of Business – Department of Business Law

A deflated air mattress rests in the corner of Airbnb’s world headquarters. It symbolizes how Airbnb allows regular, local people to earn extra income by renting out space in their homes. Yet, this symbolism fails to represent what the company has become—a unicorn receiving most of its revenue from professionals with full-time listings. That poorly folded wad of plastic exemplifies the Myth of the Sharing Economy, which has been consistently used to subvert regulation.
The Myth convinces people that the sharing economy is comprised of self-regulating platforms, which allow microentrepreneurs to utilize their excess capacity in an altruistic manner. However, the sharing economy is actually comprised of companies driven as much by market forces and failures as any taxicab company or hotel chain. The Myth possesses a simple and seductive appeal. It uses the familiar idea of sharing to make the claim that platforms are unique and should be subject to new and different regulation or no regulation at all. This Myth not only harms platform users, the environment, and the culture and diversity of communities, it has helped sharing economy platforms become powerful influencers in Silicon Valley, state legislatures, and beyond. 
While much has been written the benefits of the sharing economy and how to regulate it, this Article is the first to critique the sharing economy by exploring the intersection between narrative and regulation. It also distills lessons for regulating future innovations and demonstrates the importance of questioning rhetoric and reality in order to achieve public policy goals.

Choose how you feel, you have seven options

A great piece by Ruben Van de Ven stemming from his artwork of the same name, published on the Institute of Network Culture site. Van de Ven, in a similar vein to Will Davies, deconstructs the logic of ‘affective’ computing, sentiment analysis and their application to what has been termed the ‘attention economy’. The article does a really go job of demonstrating how the knowledge claims, and the epistemologies (perhaps ontologies too), that are at work behind these technologies are (of course) deeply political in their application. Very much worth reading! (snippet below).

 ‘Weeks ago I saw an older woman crying outside my office building as I was walking in. She was alone, and I worried she needed help. I was afraid to ask, but I set my fears aside and walked up to her. She appreciated my gesture, but said she would be fine and her husband would be along soon. With emotion enabled (Augmented Reality), I could have had far more details to help me through the situation. It would have helped me know if I should approach her. It would have also let me know how she truly felt about my talking to her.’


This is how Forest Handford, a software developer, outlines his ideal future for a technology that has emerged over the past years. It is known as emotion analysis software, emotion detection, emotion recognition or emotion analytics. One day, Hartford hopes, the software will aid in understanding the other’s genuine, sincere, yet unspoken feelings (‘how she truly felt’). Technology will guide us through a landscape of emotions, like satellite navigation technologies guide us to destinations unknown to us: we blindly trust the route that is plotted out for us. But in a world of digitized emotions, what does it mean to feel 63% surprised and 54% joyful?

Please take the time to read the whole article.