«AI will gradually change our job descriptions»

In your Sparx talk, you say that in the future it will increasingly be about developing AI-based solutions such as Google Duplex or Rachel from Soul Machines. Why is this kind of “virtual assistant” so promising?
Virtual assistants have many aspects. As personal assistants on a device, they form a new, truly powerful interface between human and machine. But that’s not all, because it automatically makes them brokers between what lies beyond the machine, namely the web with classic sources of information, the knowledge of mankind, the nonsense that fills the web, a huge number of services and – most importantly – other virtual assistants and people. They will be both a curse and a blessing, whereby more attention will probably be given to the blessing. Business is already being conducted via VAs. The point when virtual assistants negotiate with each other in a legally binding way on behalf of “their people” is not that far away. Unfortunately we are not prepared for this moment.

Outside the private sector, there are a host of good examples of how assistants support human employees in their work. Combined with media such as virtual reality, augmented reality and entirely new haptic technologies, they will simplify things for people to a degree that cannot in any way be achieved through classic solutions. This ranges from providing technical documentation for a component I am currently looking at to active consultation during brain surgery, which I can perform remotely from another continent.

And then of course there is the most interesting case for companies in monetary terms: Virtual assistants between as many customers as possible and an organisation. You don't need to beat about the bush: If they become good enough – and Rachel from Soul Machines is still far from it – entire call centres or departments entrusted with end customer communication will be reduced to a minimum workforce of just a few people, or become completely obsolete. According to an Oxford study, the time by which a vendor is completely replaced by AI is between 7 and 35 years in the future. The most likely value is less than 15 years.

Speaking of the Oxford study: What is your personal forecast – in which areas will AI have a big impact in the coming years?
AI will change things everywhere – the world of work and society in general. In an unexpected way too, for example by not using AI. The impact will be transformative at an early stage in some areas – the potential for automation is decisive. AI will initially provide support in other – complex – areas.

The areas of interest are places where society decides not to go to the technical end of the possibilities for automation. The “AI” option will hopefully provide for discussions that will redefine the “value” of some professions and activities, be it social workers, teachers, psychologists or artists and landscape gardeners. We will have to ask ourselves “What things do we accept when they come from a machine?” In this context, studies from Japan are very interesting: People don’t like robots that are too human. This creates fear. If they are more like comic figures, services are more likely to be accepted from them.

In your talk, you describe an already existing AI solution as “alarming”: GPT-3 from OpenAI. This can not only write sophisticated texts – like Hemingway would write about Harry Potter, for example – but has also been optimised so that it can generate a website from the purely vocal description of layout requests. So do those of us with a high level of training and in supposedly modern, creative professions such as Content Creator or Requirement Engineer have to worry that a machine will soon be better than us?
Yes and no. My expectation is that the job descriptions of many different roles will gradually change over the next few years and decades. In the end, new roles will emerge that will require the ability to handle AI. Low-qualified jobs will definitely disappear. And there will be completely new jobs instead. The data labelling required to train AI alone guarantees tens of thousands of human jobs worldwide. The World Economic Forum has published a study saying that, for the first time in 2020, more jobs have been created than have been lost due to AI.

And if you really want to know if your job is at risk, you can go to https://willrobotstakemyjob.com/ and browse. In addition to finding out if your own job is at risk, you can also find out which jobs are faring better. But please: Don't take it all too seriously! ?

In the software industry, you expect jobs to change, but there will be more rather than less. What professions or activities are specifically at risk of becoming superfluous through AI?
It’s impossible to say at the moment. Highly qualified jobs are far from being in danger. Jobs that have potential for automation are. All methods of working, however, will change. I think whether a job will generally become superfluous is in most cases no longer a question of technology. It is a result of negotiation within society: “Do we accept that a role is filled by a machine or do we prefer a person?” It is also a question of culture or personal attitude: “Do I accept the robot waiter or go to another restaurant?” “Do I want AI euthanasia?”… Completely new social lines will be drawn and our actions will be redefined and evaluated.

When testing software, Sapienz and SapFix from Facebook already provide solutions that can detect and even eliminate bugs. With Sketch2Code, Codota or intelliCode, there are other solutions that can circumvent a designer or take over the work of a developer, although there is still room for improvement here. You are calling for these developments to be dealt with today. Who and how do you mean specifically?
I mean all stakeholders. Performance and quality are crucial both in the development and operation of solutions. If there is something that moves a team, department or organisation forward, then I need to look closely at it, evaluate it and draw consequences from it – otherwise I’m not doing my job properly. This “something” is just around the corner.

What do you advise companies and individuals to do in order to be as prepared as possible for the upcoming opportunities and challenges of new technologies?
The availability, management and use of information are crucial. As a company and as a person, I must always have a clear picture of which innovations could benefit me and which could be dangerous to me. Today, it is no longer enough to just be informed in the professional sphere as there are many things, some from unexpected directions, which can influence my company. For example, the improved abilities of delivery robots are forcing construction companies and architects to plan new entrances. But which architect is already dealing with RPA?

All of this is difficult because although most companies should be investing here, they are not due to cost pressure, other constraints or difficulty thinking outside the box. In this case, a company should perhaps seek external advice in the form of a trend radar.

You yourself have developed the “Go To The T” app to help people improve their squash game. Will trainers be redundant in the future, as AI is much better at collecting and evaluating all the athlete data?
If I can replace a squash trainer with the app, they would be a failure and deserve it. Trainers in particular combine different roles: From a merciless slave-driver to an all-knowing mentor and shoulder to cry on. Every role is profoundly human. AI will help trainers, professional athletes and fitness enthusiasts in the shape of tools with which they can – significantly – increase their performance. We are still right at the beginning here. Many professionals in many sports don’t even have a healthy data base on which to build their training. But AI will not be able to replace a trainer for a very long time – unless the training athlete is a real AI nerd. ?

Watch the Sparx episode with Martin Luckow