Some people, including gurus like Elon Musk and Stephen Hawking, are scared by the possible applications of artificial intelligence in the military. Others see machines with AI as a threat for workers, stealing away their jobs. But a large number of other experts tend to downplay these worries, treating them a bit like superstitions for easily impressed kids.
Since a number of these supporters are managers, employed by the very same companies that are developing the AIs, one might wonder if they look so carefree and self-assured because they know what they are talking about, or just because they believe the upcoming revolution is not going to affect them personally.
Like in the tale of the guy with the pentagram and the holy water that Musk recently recalled, they believe they can summon the daemon and control it too.
Well, there's something that maybe they should consider before jumping so cheerfully on the bandwagon: it's not just low-skilled workers that are going to be affected by the rise of autonomous agents, it's also the management that is at risk. At Belfast's EnterConf conference, I recently had a mind-opening conversation with Singularity University's Nell Watson on how this could happen.
"Thanks to this combination of blockchain technologies and artificial intelligence," Watson, who is adjunct faculty within the University's Artificial Intelligence (AI) & Robotics track, told me, " we're now potentially able to create distributed autonomous organizations (DAOs), for the first time."
"What that means is that is theoretically possible to bootstrap a profitable business in, basically, a day, with a distributed team performing all kinds of different functions all over the world. That's hiring, procurement, supplies, all of that stuff."
So, how would that work in practice? The blockchain, remember, is the decentralized, secure database that records and stores every transaction that occurs in a network. It is used by the Bitcoin crypto-currency, but is not restricted to it.
It's basically an infrastructure that provides a 'layer of trust' (or makes possible a 'trustless environment, as some prefer to say, in which 'trust' is not needed, as it's pre-coded in it) on which is possible to run several applications. One of the most interesting being the so-called 'smart contracts'.
Keeping it simple, we could say that smart contracts are contracts that are made valuable, enforced and executed through digital means, without the need for intermediaries that certify the validity of the transaction or check that the task has actually been performed.
In a seminal article written in 1997, Nick Szabo used the humble vending machine as an example to illustrate the concept. Whenever you buy a product, you stipulate a contract whose terms are somehow embedded in the machine. The machine takes in coins, and via a simple mechanism dispenses change and product according to the displayed price.
In today's data-driven world, thanks to the Internet of Things and widespread connectivity, the possibilities are much greater than this. Smart contracts could allow you to do almost everything, from buying and selling a property to check workers' performance and pay them using the blockchain.
"That enables a whole new element of disruption. In theory, we don't even need to register a company anymore. You can have this entity that generates profit running in the cloud and it's completely anonymous and it's very difficult to trace who actually owns it," Watson says.
There are a number of platforms that are already being created to take advantage of that. One of them is called Ethereum, and offers the promise of "unstoppable applications without any possibility of downtime, censorship, fraud or third party interference."
Most interestingly, these platforms would not necessarily need a management. They could be run by an AI that controls all the processes, assembles automated tasks with human contribution, and provides the final service. Humans would still be involved, to a degree, but they would not run the whole show.
Think this is too far-fetched and your management job is too "creative" or smart to disappear in a snap? Think again. The Harvard Business Review recently run a fascinating article by Devin Fidler, who heads the the Workable Futures Initiative at Palo Alto's Institute for the Future, in which he discusses something called iCEO.
He describes it as a virtual management system able to manage actual work projects for a real organization. It does so, by dividing complex work into small into small individual task, in an assembly line that could be run from a dashboard. It's not just a theoretical concept: the team at IFTT created a prototype software to check if their assumption were right.
Specifically, they programmed iCEO to oversee the preparation of a 124-page research report for a Fortune 50 company. The software used workers on Amazon’s Mechanical Turk, technical analysts from oDesk and freelance writers on Elance.
It routed tasks across 23 people from around the world, to complete the project; people that, ideally, could not even know that they were dealing with a programme and not a human being. And guess what? It worked.
While Ethereum, other blockchain based platforms and the iCEO project are not necessarily interrelated, they are two good examples of how researchers are proactively working to eliminate - or at least reduce - the need of having someone at the top of the corporate pyramid.
If this could be actually implemented, it would bring along a number of issues - the most important of which is probably liability. Imagine a platform run by an AI is involved in some kind of illegal activity, or something goes wrong with a worker: who takes the blame?
"That is a question that cannot be answered at the moment legally. This is something that has agency. It is able to act and do stuff, like hire people. But, there's no legal person. There's nobody to prosecute," Watson says.
This could potentially become an even bigger problem if, as some believe, the so-called Singularity will happen and machines will become much more intelligent than humans, and autonomous in their choices.
While it's possible to encode into smart contracts the parties' obligations, coding ethics into an AI is an another thing altogether.
"In the worst-case scenario, we could create an AI agent to make profit and it would decide, "OK, here's a good way to make some money. I'm going to tweet all kinds of stuff about terrorism. The markets crash. I buy lots of stock. People realize within a couple of moments that it was nonsense. The markets recover. I sell stock," the expert adds.
It would make sense, from a rational, machine-like point of view. But this, of course, is only part of the story. On the bright side, DAO could be used, for instance, to create new business models not based upon profit, but around public service.
"Traditionally, public service organizations are limited in the scope of their activities by personnel processes, for example. Of course, DAOs are able to sidestep a lot of those costs.Interestingly also, you could have a distributed autonomous organization that makes profit and distributes it to human shareholders."
"You could have a kind of a social welfare system based on robot businesses making profit in perfect competition with each other in the cloud distributing those profits to actual human beings, which might be an alternative to stuff like benefits," Watson maintains.
It might sound a bit like Sci-Fi at the moment, and the whole thing is in its infancy. But the technical feasibility of some of these platforms (the non-conscious ones), still as rudimentary as it might look, has already been demonstrated.
So it might be worthwhile to begin a debate on the subject while we're still in an early phase, and we're still able to choose in which direction to funnel the change.
