Forget the tech, managing people is the hardest part of introducing ‘AI’ into your business
TL;DR: The people challenge of introducing ‘AI’ into a business is often underestimated compared to the technology challenge. This is a guide to the human processes and pitfalls to consider when introducing and implementing AI-powered tech.
AI — it’s the big buzzword. Depending who you ask, it’s either going to save the faltering economy or it’s going to steal all our jobs. Or maybe both.
Certainly machines are doing some amazing things for businesses:
- writing marketing emails automatically, based on a few bullet points
- automating customer service
- dramatically reducing the cost of content creation
- handling sales leads more efficiently
But this focus on the wow-factor of the technology also obscures the role of people in it.
The recent A-level fiasco in England has shown us that no algorithm is neutral, they are all designed by humans to deliver outcomes that meet human needs.
Likewise, how you implement AI-powered tech into your business is as much a people problem as it is a technological one.
As with any business change, the introduction of AI-powered technology will have potential people blockers. There are those in the business with vested interests, who may resist change, sometimes very strongly. There will also be people who have genuine worries about what it means for individual job satisfaction or status.
Not to mention the real issues of embedding new processes and workflows into your organisation.
How you implement AI-powered tech into your business is as much a people problem as it is a technological one.
I’ve seen all of these challenges at play at one time or another and these human factors need to be managed as closely, if not more so, than the technology itself.
You’re not just changing a business, you’re changing social structures.
Managing The Human in AI
Luckily, this is not a new problem. There has been a lot of work done on why major change initiatives — such as the introduction of machine learning to a business — can generate only lukewarm results, or even fail.
Professor John Kotter at Harvard Business School created a model back in 1996 is still recognised as one of the best.
Kotter’s main insight was that transformation is a process, not an event. It advances through set stages that build on each other. If you try to shortcut the process, to skip stages, it won’t work.
He identified eight critical success factors — from establishing a sense of extraordinary urgency, to creating short-term wins, to changing the culture longer term.
This framework is well worth taking a look at. It’s as valuable and relevant today as when it was written 25 years ago and is a great road-map for implementing AI-powered tech into your business.
One message won’t fit all
As you can see, Kotter’s framework is very much aimed at the leadership of a business.
Senior management will be highly motivated by appeals to business opportunities or threats. After all, they are also likely to be shareholders or have their remuneration based on the company’s financial success. But messaging and approach may need to vary for other people in your business.
For many, the motivating factor is likely to be personal. Their work will become more interesting. They will be more valuable to the business. The retraining or knowledge they will gain will make them more valuable in the market.
It’s worth thinking through the issues from many perspectives when you build your communication plan.
Use understandable terms
One of the weak points of AI/machine-learning is the massive amount of jargon used by its practitioners.
It is a new area and one that is driven by scientists and academia, so it’s not surprising that a lot of technical terms crop up. When experts in any field talk to one another, they want to shorthand complicated ideas. Which is great for them, but doesn’t work for us.
People in your organisation need to not only understand what is being asked of them, but also how to communicate it to others. There is no room for error. Ambiguity leads to confusion, confusion kills campaigns.
So, find a way to make that jargon intelligible. Be ruthless.
(“Clear writing and clear speaking lead to clear thinking” is one of our mottos at Woolf)
Involvement with respect
Many studies have shown that imposing change from above or outside is are often ineffective and met with more resistance.
If people are denied the satisfaction of being part of the solution, they will often gain satisfaction by resisting change.
Involvement or participation of the teams affected by the change will also lead to better results. There will be all sorts of implementational problems along the way, which staff will have the answers to. (“None of us are as smart as all of us”.)
Engaging with people is key — before, during and after the change.
Depending on the size of your business, this might take the form of roadshows, forums, calls or in-person meetings to discuss the changes. During these sessions, encourage interest in the project right from the start and throughout. Demonstrate that people’s input is not only sought but, where appropriate, acted on.
If people are denied the satisfaction of being part of the solution, they will often gain satisfaction by resisting change.
For example, machine learning often relies on data. Often the data exists but is spread out across the business in silos. So, you will be reliant on many people or departments coming together to agree a solution.
The key here is not to use ‘participation’ cynically as a way to manipulate people into doing what you want. You and your managers need to really mean it and live it.
Retraining
People will almost certainly need retraining when you introduce AI-powered tech into your business.
Automating a part of a someone’s job doesn’t make the other parts unnecessary, in fact it usually makes them more important and increases their economic value.
In the world of design, for example, there used to be people whose job was to cut and paste bits of paper to create artwork. Those roles disappeared with the advent of computer-aided design. Design departments themselves massively increased as it became more affordable to ‘design’ almost everything. And those who made the leap, learned to use the software and operate the machines, were better paid and did more interesting work, focusing more on creativity and innovation than tedious manual tasks.
A very similar thing is happening now as machine learning is introduced into businesses. While the machines can do the heavy lifting, it will still be human beings who set the direction, interpret and determine the outputs as well as program the machines themselves
But all of this does require flexibility and the willingness to change. Which leads us on to the last point…
It’s not for everyone
The reality is that not everyone is going to get on board with a programme like this. You need to do what you can to bring people with you — but it doesn’t have to be everyone, and it probably won’t be.
For some, this next phase of the business might not suit them and it might be absolutely the time to move on. And that’s fine.
If this, or any of the problems described here, chime with you, please leave a comment or get in touch
Please clap or share this article if you have enjoyed it as it helps other people find it.
If you’d like to be on my (100% spam-free) mailing list to be updated when a new article is published, you can sign up here.
