Meet QIARK, a new AI system that helps organisations to better understand what people think – and which could have some interesting use cases in the law, especially around delivering better insights into innovation and KM. And, it could even be of use to management teams seeking to decide the strategic direction of a legal business – or any business for that matter. It could also perhaps help with making actual legal decisions.
Artificial Lawyer recently caught up with the founder, Mark Ricketts. First let’s understand what QIARK, which has been developed by Oxfordshire-based company, ImpactRI, actually does.
‘We believe that it is possible for people to collaborate in ways that can outperform the most powerful computers, and do so in ways that harvests the full extent of human endeavour, for the common good of commerce and the people it serves,’ is how the company officially puts it.
Or, as Ricketts explained:
- You decide a question (or series of questions) that you want an answer to, this could be anything, for example what do partners at the firm see as the main innovation challenge?
- People reply in natural language, explaining their thoughts in their own words and not needing to fit into any particular format or style. I.e. they just say what they think.
- The system, which has a variety of NLP, ML and pattern spotting neural network software at work reads everything, looks for similarities between all the expressed thoughts and then groups them into logical collections of ideas. I.e. let’s say that 10 people have the same basic need but they express it in 10 different ways, the system will still see the connections and group them.
- It shows you the most popular overall ideas, allowing you to understand what the group really thinks and what really is the key thing people are focused on. And this should help avoid buzzwords and getting trapped in reductive polling of over-simplistic statements. I.e. you get a much more subtle output.
- You can leave it at that, or then move onto an adjudication phase that allows a further level of triage.
‘This is sorting out ideas. The data can be from 1,000s of voices. But how can [organisations] deal with 1,000s of ideas to solve a problem? You need to find the key points. You need to sort out the opinions,’ Ricketts explained.
He adds that they have had serious interest already, including Members of Parliament who want to know what their constituents really want and think. He adds that the beauty of this system is that it’s not a poll where the answers are given first and you have to vote on it. For example as happened with Brexit, i.e. people had to vote on what was a very opaque statement and then the numbers were added up.
With QIARK the question would be more like: ‘What would you like to do about the UK’s relationship with the EU?’ And then the system would sort out thousands of ideas expressed in natural language. You could then see what really were the key issues and perhaps that would show that what you expected to be the answer was very different and not so binary in nature.
But what uses does this have?
Clearly in politics….including the politics of business strategy it has a huge potential use. For example, Artificial Lawyer’s alter-ego, Tromans Consulting, is often asked to help legal services businesses and tech companies to better understand their strategic and/or innovation needs. A tool such as this, with a series of appropriate questions could quite rapidly help to triage what the organisation really wanted. I.e. rather than people saying something fairly oblique such as ‘we want AI’, the questions would draw out more complex responses, leading to a clear understanding of what really was the key issue. This could potentially reduce the need for face to face consultations with people in an organisation to find out their real issues. (Whoops….looks like I’ve put myself out of a job….)
But, perhaps there are many other uses. What if a law firm wanted to gather feedback on a tech tool it was using? Or perhaps a group of lawyers wanted to decide on a legal issue, perhaps a form of wording, or an approach to a precedent? Could this be used by a group of lawyers trying to decide the best strategy to handle a legal matter or agree on how to apply the law?
Could perhaps jury members in a common law court use it to express their views on a case, or an aspect of the case – and would this be useful to the judge? Could it be used in a private arbitration to come to a balanced solution?
The more you look at it the more you start to see use cases, and AL has only just started to think about it. A firm that had some time to test it and really explore use cases could likely come up with many new applications for QIARK.
One thing that this does in particular is drive the idea that group knowledge could be better than that of one or two experts. I.e. fundamentally it’s putting faith in ‘the wisdom of crowds’ – that if you can extract sufficient insight from a relevant group about a problem they can come up with a meaningful answer.
Of course, there is a counter-argument, which is that just because X number of people believe Y, that does not make it correct. One single well-informed expert may have better insight than a dozen people who may not have spent time considering a certain issue. (So, maybe consultants do still have a job…?)
Fundamentally the benefit here is that the organisation asking the questions will really know now what people think – as QIARK can capture their expressed thoughts, not just depend on a parametric poll, as Ricketts notes.
Ricketts adds that you can keep asking additional questions to help you to refine insights. The system will also show the statistical balance of the outcomes, so you can see which ideas are most supported and which ones are not.
Is this going to take off? As of now QIARK is seeing a wide range of interest, from MPs in the UK as noted above, to the Danish Government, to organisations such as St John’s Ambulance.
It’s certainly one of those tools where any innovation team and/or management team could likely get some value from it. The trick is of course knowing what questions to ask.
Check out the video below for more info.