AI is Brexit

  • 05 Mar 2019

If I ask the Twitter universe about Brexit, a subject that will clearly touch us all, I get some pretty strong opinions. This stands to reason because (1) it’s an emotive topic (2) often these are people already actively involved in the Twitter debate (I imagine the rest have better things to do than respond to my Tweets).

Nevertheless, this can sway my view of what the world thinks about Brexit.

It’s with this in mind that I read the latest McKinsey report on AI Adoption (mckinsey-ai-adoption) across the globe and compare it with my 1st hand experience of talking to CEOs in the UK. Like Brexit, they know AI is on their doorstep but it’s complex, potentially costly and it seems possible to kick the can down the road a while longer … until it’s too late and you’re out of business (or Europe).

But, of the people who are making some inroads in the AI space, what are they up to and how are they getting on?

Quick Summary;

AI adoption is expanding at a pretty decent pace. Close to 50% of respondents say they have embedded at least one AI capability in to their business, up from 20% the year before and a further 30% are piloting the use of AI. A minority of 21% said that they have embedded AI in to multiple business units. To give some perspective, the proportion of digital spend is still small, c10% of total spend, but it’s set to grow significantly.

So, what are the reported effects of adoption so far? McKinsey say that results from AI use have been “moderate or significant”, whatever that means. Still, it’s clear that there’s a significant majority of companies reporting that they are partly of fully on their AI journey.

IMO; As Director of a data consultancy, of course I am fully on board the AI bus and supportive of the capabilities and significance of AI. But, in our conversations with CEOs and other data leaders I would reduce the above numbers by at least 50%, probably more. Where I do agree with the report is in the findings related to the barriers to adoption below.

Top 5 AI Capabilities;

There are multiple use cases for AI. Bare in mind that the type of adoption is relative to the industry but the top 5 are;

  1. Robotic Process Automation
  2. Machine Learning
  3. Conversational Interfaces
  4. Computer Vision
  5. (in fact 5, 6 & 7) are all related to NLP (text and speech).

IMO: We have a biased view because we are more often engaged in retail, high tech, media and B2C consumer products and service industries than, say, manufacturing or assembly. Therefore, points 2-5 are more typical projects for us to advise and deliver on. However, we often find detailed use cases thin on the ground and their prioritising somewhat random. This really is an essential part of successful planning, especially in untried fields like AI.

So who’s using them?

Of course, different functions are utilising AI in different ways and most parts of any business can engage with AI. However, with a few exceptions (namely Financial Services in Risk and Automotive & Assembly in Manufacturing), the functions making most use of AI are;

  1. Service Operations
  2. Product Development
  3. Sales & Marketing

IMO: Sales & Marketing is the function most readily talked about in relation to AI, in terms of making use of customer data. However, this can be the most difficult as the quantity and quality of data is often questionable. If relevant to your business, you might want to consider the areas of pricing and stock control to be more immediately impactful on your returns.

And the barriers to AI Adoption?

Without harping back to the Brexit metaphor, people often have plenty to say about their AI use without actually having a strategy or much conviction about what they are saying. The fact is that most businesses that talk about having AI capabilities are really talking about data science at best.

Nevertheless, there are genuine reasons for lack of adoption of AI (you can read my full view of these elsewhere) but according to McKinsey’s (and I fully agree), the big 2 are;

  1. Lack of clear strategy
  2. Lack of talent

These are closely followed by tier 2 barriers;

  1. Functional silos
  1. Lack of leader’s ownership & commitment

The common thread … is you (or other people) with some structural issues (or challenges if you prefer) thrown in. These must be addressed early and continuously throughout the project … definitely not ignored. Like many things in life, just throwing money at the problem (and that includes simply outsourcing to a consultancy or ‘hiring a head of’) is not a going to work.

IMO: You’ll notice that none of the top 4 barriers are related to technology, so stop buying it until you’ve addressed at least 3 of the other 4. You should always tackle 4, 1, 2 in that order because without these ducks lined up, you will fail.

It’s not going away. AI will affect everyone in the end, from this generation and the next, just like Brexit. It’s no good sitting back and ignoring it. IMO, take action.

  • AI