New work constructs in the intelligent age
In June 2016, I wrote in this column of the changing nature of employment contracts within a firm and how we might soon see a marketplace for information technology services. In that column, I posited that the contracting cost around buying and selling software and business services is plummeting, given the ease with which humans can connect to the internet’s ‘cloud’ and to complete parsed-out pieces of work sent to a ‘crowdsourcing’ marketplace so that the completed parsed tasks can be later assembled into a whole.
This reduces the need for a services firm to act as a tight unit of organized labour, which calls into question whether services firms need to have as many regular employees as they do now. The future may be one where services firms become simply a platform for buyers and sellers to collaborate. The platform will itself only provide a few signature quality control and pricing processes which buyers and sellers in the services market will tacitly agree to in order to use, deliver and pay for services.
Building on this construct, I wrote later in the year that Wipro Ltd’s acquisition of Appirio Inc. was appropriate, largely because the latter firm owned the two largest ‘crowdsourcing’ platforms available in the computer programming world, called ‘Top Coder’ and ‘Cloud Spokes’ which together act as a platform for well over a million computer programmers. Platforms such as these can be used both to collaborate internally as well as to trade services between buyers and sellers.
It appears as if this phenomenon of ‘crowdsourcing’ is gaining currency in many parts of the artificial intelligence (AI) world as well. Last week, I mentioned Open Clinical, which is a crowdsourcing platform for collaboration in the world of medical practice, and a repository of expert knowledge in bio-informatics for AI. Other, less automated platforms seek to find top consultants or experts, who can consult in bite-size pieces of an hour by providing expert advice to buyers who are interested in gaining from these experts’ views on a variety of topics. These knowledge stores and platforms are increasingly being used to drive economic engines, fuelled by those who contribute to them and those who use them.
There are other, less specialized areas where new AI is creating new job opportunities rather than destroying them, and very many of these opportunities seem to be following the platform or marketplace construct.
In a recent edition of The Economist, the magazine highlights many employment opportunities that this sort of crowdsourcing marketplace work construct has already created, and highlights some of these areas, especially ones where specialized knowledge isn’t really needed, but a human eye or brain is still required to perform ‘micro tasks’ that can make sense of large data stores for machines running AI programmes. Facebook, for instance, has faced criticism for allowing violent and otherwise objectionable video content being posted to its site and has had to increase the number of workers who view and police the content to quickly remove posts that can cause outrage.
AI, without reliable data, is dead on arrival, and humans can play an extremely important task in reliably labelling these data for a variety of uses that AI seeks to accomplish: from labelling traffic signs or road hazards such as pedestrians or potholes that can be used by the driverless cars of the future to transcribing bits of audio that can be the user interface as AI applications move from today’s typing reliant devices to tomorrow’s speech-enabled engines. The magazine claims that such labelling work already keeps thousands of people busy.
It also talks of experiments where ‘virtual companies’ are assembled by software for specific projects, only to disband when the project is done. This is not a new idea; India’s enormous ‘unorganized sector’ sees this sort of organization come together every day in its physical market places where individual actors in a network convene each morning, say to sort and grade vegetables as they come into town, finance the many push-cart sellers of these vegetables, and then collect the proceeds later in the same day—only to wake up and start this process all over the next day, often using different individual actors in the network.
Such manual work is also important when the AI behemoths compete with one another. Google, for instance, has recently announced a joint venture with Wal-Mart Stores Inc., as both firms seek to counter Amazon’s relentless march in online retail. Amazon which has over three quarters of market share in the online retail space, has bought grocer Whole Foods to boot, and one of the many alliances that Google’s own competing e-tailing service needs to counter the retail behemoth is this tie-up. It seeks to marry Google’s voice-based Google Assistant to Wal-Mart’s grocery shelves. It hopes that users will simply dictate their grocery shopping needs to Google’s Assistant so that the groceries can either be delivered home, or picked up outside a Wal-Mart store on a hungry commuter’s ride back home from work.
But the reliance on voice instead of a typed-up list is fraught with problems, since grocery shopping throws up a variety of options in size, type, colour and quantity that are more easily rendered onto a graphical and keyboard-based online platform than delivered by a voice exchange. For instance, when you ask for soap, you are faced with a variety of voice-based questions such as: Which type of soap? Which brand? What size? How many? And so on, which can be a frustrating experience when compared to simply looking up these options and clicking a ‘select’ button on an online keyboard. There is no doubt that Google will need to rely heavily on manual ‘labellers’ of voice-based options in order to make the experience smoother for its customers.
These new marketplaces will create more work for humans as the Intelligent Age takes off, and a lot of it will not require a PhD in computer science.
Siddharth Pai is a world-renowned technology consultant who has personally led over $20 billion in complex, first-of-a-kind outsourcing transactions.