our views and our knowledge in analytics and other releveant topics

our blogs

Thinking for yourself - the dangers of going too far too soon with AI


In his 1987 novel Dirk Gently and the Holistic Detective Agency, Douglas Adams introduces a character called an “Electric Monk”.  This is a labour-saving robot/humanoid.  It operates a bit like a dishwasher - saving the owner time by washing the dishes for you, or a microwave cooks food for you.  But the labour that this Electric Monk saves was thinking.*  It does your thinking for you.  So that you don’t have to.

When I read the book when it first came out (it was one of the first contemporary newly-published novels I read, having recently devoured the Hitchhiker’s series), I thought this was a curious characteristic creation, even for one of the most inventive novelists of his generation.  I didn’t really see the point of it.  Surely the point of all the physically labour-saving devices that the 1980s created was to give the individual person more time to think for themselves, and so do whatever they decided they wanted to do.  So, to outsource the very process that made you unique – how you think – would be pointless, and stupid.  And, as a young teenager, I enjoyed thinking – I believe it’s what teenagers do best, for better or worse - so this felt doubly odd.

What is the point of not thinking?

Fast forward 30 years, and at the beginning of this year, I read a blog, by a Professor, no less, about a major international Business Intelligence conference.  There, he met with a range of executives about how their organisations were planning to leap forward from the drudgery of having analysts to understand their business – they had missed the opportunities afforded by having an insight and analytics team.  This meant that they were now behind the competition who had invested in this relatively basic analytical – one could call it a “thinking” – function.  The blog says:

“In fact, several companies (I have promised not to name them) said they were hoping to ‘leapfrog’ over traditional business intelligence and analytics capabilities and go directly into AI-related environments. As one technology executive from a software company put it, ‘We’ve had enough bar charts. Nobody has time to digest them anyway. We want our analytics to tell people what to do.’”

In other words, they are looking to outsource their entire corporate intelligence and performance analysis function to the real-life business equivalent of the Electric Monk.

I think there is a lot of value in predictive analytics and Artificial Intelligence (AI).  Indeed, I think the traditional BI teams try and do too much by hand, and can often be reluctant to “let go” of activities.  One such example is algorithmic attribution, where the weighting of the performance of different acquisition channels is done by an algorithm – a computer – rather than by the business.  Given there are so many stakeholders involved in the outputs of attribution, handing this over to a computer, free from any internal political bias, is a distinct advantage, and can significantly improve performance, provided, of course, that everyone can agree to implement the results.

And that last part is why I am concerned with the idea of companies entirely outsourcing the insight and analytics function to AI.  In order to get the best results from such a piece of AI, you need to have a data-driven culture, where your teams embrace the findings of insight, and look to improve their business performance on the back of it.  It’s effectively a prerequisite for successful usage of insight; without it, business processes, behaviour and attitudes won’t change, because people will question the results, or, worse still, simply ignore what the data are telling them.  Or, to put it another way, it means the business hasn’t learnt to think for itself.

The idea of missing this step on the path to insight maturity is therefore at best risky, and at worse, self-defeating; without the internal “self-knowledge” of what the business can or should achieve, how can you make sure that your software and predictive analytics tools are doing the right thing, or are focused on the most important business question?  And some of these very alarm bells are included in the quote above – if your teams don’t even have enough time to understand some bar charts, what hope do you have of becoming data driven and using AI properly?  This has all the hallmarks of very expensive technology being implemented without thinking about the true business requirements, and then never being used.

So, the development of AI is not the end of analytics.  It’s where having good analytics and insight capabilities is the very starting point, the baseline of your knowledge, on which AI can build and enhance your business performance.

* In fact, in the book, the Electric Monk humanoid was designed to do your religion for you.  But the one in the novel was a malfunctioning one, so it thought and believed just about everything and anything.  So it did in fact do the thinking for you.  This blog is too short to go into the differences between belief and insight, particularly in relation to companies; it probably needs a whole book!