The Need For White Box AI In Legal Work – Ayfie’s Johannes Stiehler

As 2020 kicks in, tech companies have moved into the new year with fresh goals and ambitions, many eyeing a bigger stake in the steadily growing legal tech market.

Among the crowd, text analysis company ayfie can be found, and perched in its offices, with his head likely buried in an innovation challenge, is Johannes Stiehler, Chief Technology Officer.

For Stiehler and colleagues, the year has started well and the team is keen to build on its improved text analyzer, Inspector for Relativity, which is used for e-discovery and investigations work, and which to put it simply, helps to eliminate the need for human review on most of the non-responsive portion of a document collection, among other things.

But, Stiehler also has his mind focused on other challenges.

White Box AI Vs Black Box

One particular area that Stiehler wants to nail in 2020 is to make their solutions more ‘white box’, that is to say ‘more defensible, predictable and easy to understand for clients’, as opposed to the more ‘black box’ scenario we are currently in when it comes to machine learning and algorithmic decision-making.

Stiehler believes that clients for the most part don’t really understand machine learning processes, but they do understand the need to be able to follow the reasoning of the software to a certain degree.

As an example, he said generally people don’t understand how Google’s image search recognises images, but they are fine with it because the results are good, so they just use it. 

But in the legal space we can’t afford the luxury of not understanding why something happens because you might have to defend it in court, or you might have to understand why it works in a certain way and potentially correct it using human input,’ he added

To that extent, ayfie has invested a lot of time in working on this challenge, he added, and will continue to do so across 2020.

‘I think that will be a key area where we are looking to invest resources and innovate. And this may mean, for instance, sometimes using an algorithm that is not maybe even optimal or not as good as another algorithm, but it is far more predictable or accessible to human reasoning,’ he explained.

And that is a first for Artificial Lawyer. Over the years this news site has spoken to many people in the field of legal AI and text analysis, and Stiehler is the first to state he believes vendors might need to explicitly balance their software in favour of transparency over the need for optimum output.

Innovation: What Is It? How Do You Sustain It? Is It Cutting Costs?

Artificial Lawyer also asked ayfie’s CTO his views on three key questions about innovation in the legal sector:

What is innovation?

It is generally the renewal of products, services, processes or concepts. The renewal efforts are driven by the need to deliver better outcomes, or deliver the best results more quickly or at lower costs.

How do you sustain innovation?

Some of the innovation in ayfie is driven by our customers, and I think that’s a litmus test for any company.

On the other hand, we also have innovation as a constant factor in our lives, driven by the creativity of every single ayfie employee, so people are constantly finding new solutions.

We might store those solutions for future use, or discard them altogether. Generally, the people who work in ayfie – and that is also the reason why we hire them – are very curious to find solutions to existing problems or to even find or invent new technology.

Is the use of legal technology reducing the cost of delivering legal services? 

Ideally, technology will reduce the cost of legal services. It depends on how technology is embedded into business processes. For example, you can get through due diligence by using automated contract analysis, but [there may be] training for the solution that takes such a long time that you actually don’t save any costs at all.

By Irene Madongo