Management consulting helps thousands of small businesses to get off the ground and big businesses to thrive. I would say that – it’s what I do and I’m proud of it. People trust our experience, skill and intelligence to help them, but the growth of Artificial Intelligence (AI) may pose a novel threat to the established order.
We are familiar with AI in automation of vehicles and manufacturing, analysis of climate models and for rapid development of drugs or therapies in healthcare, but it can do so much more.
I listened today to Tim Ferriss in discussion with Eric Schmidt on The Promises and Perils of AI (amongst other things) and it woke me up to the alignment of what I do as a consultant and coach and what AI can do, but so much better and faster. It sounds scary because fundamentally, decision support by a consultant or AI follows a broadly similar path.
When working with clients I aim to help them to make good decisions that can be acted upon with confidence for a positive outcome. This often relies on building hypotheses and testing them to eliminate the weaker ones. AI decision support, often referred to as optimisation, does similar stuff.
For entrepreneurs, decision making can be tough. When a situation arises, there is a temptation to make no decision and let things ride, or maybe make an emotional choice that is heavily biased to the instincts of the decision-maker. The first is troubling as it denies progress; the second may work but can be difficult to explain or get buy-in from others with a different perspective. Either way, building confidence in what happens next is tricky.
Whichever path is taken, the decision process should be rational and focused on delivering a definite and beneficial outcome. The process requires a well-understood context, paired with a trusted outlook (or range of outlooks) to allow the development of several hypotheses. By testing these to eliminate the worst and compare the good, it is possible to find the best. As a classic consulting approach, this is entirely consistent with an AI optimisation system.
In practical terms, and having consulted like this for many years, I know that it can be an intensive task to reach an ‘optimum’ recommended path. Worse still, if any of the variables moves, then the choice may no longer be ideal so the process must reset.
The beauty of AI is the way it works with lots of data, looks at many possible pathways and almost instantly works out a preferred option. Even better, if anything changes, the computing power allows that to be done again and again before you know it. One up to the AI team.
So am I and my peers about to be marginalised? I think not, or at least not for a while. Whilst the theory is great, there is a way to go before AI solutions become ubiquitous. They are currently well suited to a large-scale dynamic problem where there is a lot of data that can be gathered and processed consistently. This tends to mean they are large, take significant time to develop and verify before they can be used in practice, and therefore are very expensive. Moore’s law tells us that computing power will rise, and experience of building AI solutions will shorten lead times, so we can’t afford to be complacent.
My conclusion from the Eric Schmidt interview was that the approaches are similar (even if the computing power is vastly different) and there is certainly room for consultants to work alongside or with AI for many years yet.
What we must do is to recognise the overlaps and work hard on the softer elements that make us valuable. We can mobilise quickly and we can work within a pareto-style selection of the key data to produce a limited but relevant set of hypotheses to evaluate. We will develop recommendations that are very good, but not necessarily the true optimum. We will also do this quickly, or at least far quicker than it takes to develop and test the AI algorithms.
The key is to recognise the value gap between a consultant’s fast, relatively inexpensive and very good recommendation versus AI’s optimum, dynamic, expensive and initially slow solution. It’s horses for courses, and the consultant’s nimble approach is perhaps best suited to the smaller companies and start-ups where the passion of founders, moderated by some controlled processes and experience brought by the consultant, can combine to great effect.
In the long term, AI will get similarly nimble, cheaper and more responsive. As consultants, we will move from the hard yards of generating the recommendations towards validating them, providing the intellectual rigour and helping the entrepreneurs to put them into practice. In short, we will have to get smarter and let the AI systems do all the heavy lifting for us.
So no, we are not done for just yet!