Manufacturing companies, providers of software and sensor technology, R&D departments: to make the step to smart industry, everyone is exploring the possibilities of artificial intelligence (AI). Algorithms should contribute to more efficient production processes, more quality, lower costs and increased safety, but also bring new legal issues. This requires increased alertness from entrepreneurs when entering into agreements in this regard.
Datastream
YP Your Partnerr in Drachten is one such company that is primarily engaged in collecting, bringing together and processing (sensor) data. The central management of those data streams is done in its own software platform C.A.R.S, with which machines, devices and processes can be remotely monitored, operated and automated. Currently, YP Your Partner is working on the transition from C.A.R.S 7 to 8 - and in it AI should also find its place. ‘Now if a threshold is exceeded, an alarm goes off. That has to be smarter: you don't just want to receive the notification that the pressure in the boiler is exceeded, but also context information. What is the installation date of the boiler, what is its serial and type number, et cetera. Only then can you plan your maintenance more efficiently,’ says director Theun Prins.
Convert analog data
According to Prins, YP Your Partner has been successfully measuring data streams for 30 years and also has the tools to provide context. The big challenge remains getting the system filled. ‘Now, a lot of information about installations, for example, is still available in PDF form and not usable digitally. If you want to be able to use that analog data, they have to be converted into data suitable for algorithms.’ In this regard, Prins points to the BIM (Building Information Model) in construction, in which involved parties enter and share all data around installations, materials and the like. ‘We need that kind of system for industry as well. So far, only small pieces of our knowledge are used to compare or interact with installations. That concerns compact algorithms, not complex systems. There is no business case for installed base yet, and that is a prerequisite for meaningful application of AI.’
Decision Support
The added value of AI is clear: working faster and more efficiently at lower costs, with predictive maintenance as the magic word. ‘Many companies are now still in the phase of static and manual rather than interactive reporting. We want to move to fully predictive automated systems. This requires algorithms that generate decision-supporting information based on concrete queries to systems,’ states Prins. ‘That if, for example, a pump breaks down three times in a row, the software will ask whether it is not wise to replace that pump. So not only signaling, but also advising and then automatically ordering a new pump as soon as the number of failures exceeds the cost price of the pump. We as humans still have to get a feel for that.’
Chain of parties
Not only that, blindly relying on algorithms also has legal consequences, Prins knows. ‘Suppose you have implemented such an algorithm and it has worked well ten times. Ten times the pump is ordered, but the eleventh time a mega-order suddenly goes out the door. So who is responsible for that: the one who developed the algorithm or the one who applies it? And how do you then deal with that? Such a chain of parties - ranging from owner of the machine, provider of the intelligent software, developer of the algorithm to supplier of hardware and sensors - can complicate the application of AI, confirms André Kamps, independent ict and privacy lawyer at Kamps Juridisch Advies. ’If YP has devised an algorithm and built it into an application, customers sometimes apply it in completely different processes. In that case, it is good to carefully scope out in a user agreement what you are going to use the application for and record this. And further, make separate agreements about other applications, so you don't get any surprises,‘ he says.
disagreement
In addition to the question of who is liable for what piece, the lack of written agreements can also lead to disagreements over IP or revenues, for example. Prins illustrates this with a practical example. ‘Suppose we have developed an algorithm for a machine builder in a certain domain and another partner discovers that he can also apply it for his domain. Who does the IP then rest with? And how do you deal with the resulting earnings? So far, we are generally getting by with that. If a user earns solidly through that stroke of genius, we charge a predetermined percentage of the profits. But legally we haven't been able to get this right yet. We adjust agreements based on advancing insight.’ For software usage and resale rights, YP has already defined three types of partnerships. ‘Then it is established what a customer can and may do with our software. We must now regulate AI in a similar way.’
Appropriate agreements
The need for appropriate agreements will be at least as great in the case of algorithms, agrees Kamps. ‘How do you deal with confidential data? If a party wants to terminate the agreement, can he still do anything with the data? What about the rights if a party chooses to continue with the competitor? The issue of continuity is also important: what if a party fails to honor agreements or collapses? Is the system then easy to replace? Has an exit strategy been agreed at all? These are all issues that you need to take stock of in advance and adjust your business model accordingly. Other legal aspects that deserve attention are the processing of personal data (for example, when measuring the productivity of operators) and pricing in the chain.
Increasing unpredictability
As enterprise software systems become more interconnected, the use of algorithms becomes more unpredictable. And as algorithms become increasingly complex and smart/self-learning, their output is also more difficult to predict. ‘This complicates agreements on liability. In the Civil Code, damages are linked to the product or possessor; the question is whether this is applicable to self-learning algorithms,’ argues Kamps. ‘As an entrepreneur, if you opt for algorithms, you must therefore think about strict liability and insure yourself for it. However, insurers still have little experience with misinterpreted or misapplied algorithms and are probably still hesitant. Perhaps soon, parallel to the supplementary cyber policy, an information expert will assess the algorithm for acceptance in the insurance.’
Linking sources
Leading up to more intensive use of AI, Prins wants to take a number of steps with YP. ‘Firstly, linking all sources of information, virtualizing physical reality even better, and finally far-reaching human integration, which is facilitated by cell phone and interconnectivity,’ he says. ‘People need to input data and provide disciplined cause feedback, which is essential for root cause analysis. That data flow is also needed to take AI to the level where automated processes are possible. From data to action, fully automated.’
This article was published in Link Magazine from October 2019.