A leader’s guide to getting AI right

Companies are currently grappling with how to use AI, and results vary. At times it can feel like the blind are leading the blind.

As you watch leadership in your organization chart a path to engage with AI, what can you do to ensure that your company doesn’t get it completely wrong?

1. Educate yourself

To contribute to any discussions around the use of AI in your organization, you have to be educated. That education requires a few components. You should certainly be aware of the ongoing conversations that are happening broadly in the business press. But, most of the people with a platform to speak to mainstream and social media have a viewpoint and/or product they want to push. You have to take what you read with a grain of salt.

It is important to dig in and play around with some of the tools, especially since they are constantly evolving. What you were capable of doing with AI tools six months ago may be a poor indicator of what you can do with the leading-edge tools now. Engage with tools yourself to get a feel for what they are capable. Try to carry out a work-related task assisted with AI to see what the tools do well and what you can’t. Just because the tools did a poor job three months ago doesn’t mean that they won’t be better when you try again.

2. AI forward and AI responsible

One of my colleagues at the University of Texas, Julie Schell, characterized good AI strategy as being AI forward and AI responsible. The idea is that the best approach to AI is to be open to what it can do to enhance your business. Explore the ways can it simplify work processes, engage better with customers, or provide a partner to develop ideas.

As you go through this process, remember that no valuable tool is truly generic. What will make AI helpful for your organization is not the presence of a bland platform. There need to be ways that it will help you do something that you can’t do effectively already. It may not be up to you to determine how AI can make something easier or better, but your company’s strategy has to involve some process for finding good ways to use AI tools before investing in them.

At the same time, be responsible with your people, resources, and data. Don’t waste people’s time with tools that actually make the workflows harder. Resist the temptation to overpay for access to models and tools. A year ago, companies were building AI applications and charging users hefty fees to use them. With improvements in AI as a tool-builder, many of these same applications can be built by users at a fraction of the cost. That doesn’t mean you should never work with a developer, but be careful not to get locked into long-term contracts when prices for many tools are likely to go down. And make sure that you’re not giving away any of your sensitive data. Most companies protect your data when you buy an enterprise version of their models, but read the fine print.

When you look at your company’s strategy, make sure it is both forward and responsible. Leaning too heavily on either pole means you’re either going to get left behind or do something foolish.

3. Be unevenly distributed

As the author William Gibson said, “The future is already here—it’s just not evenly distributed.” Gibson meant that quote to apply to society, but it isn’t a bad characterization of effective AI strategy in many organizations.

You should always have some groups that are focused on the leading edge of technology and understanding what can be done. If the organization is large enough, this group may even have access to more tools than everyone else. This group aims to find the future AI tools for the organization. 

The bulk of the organization should be using tools that are well understood and aimed at particular aspects of workflow. It is important that the organization measure the uptake of AI tools to determine both whether they are being optimally deployed and whether the usage justifies the cost of the tools. And, of course, there will always be a few people who resist the latest tools.

Ultimately, organizations must find a profitable approach to incorporating AI. Certainly, there has to be a little investment up front to get up to speed using AI. But, it is possible to continue spending too much and getting too little out of AI. Each of us needs to monitor what is happening in our organizations to make sure that we rapidly reach a point where we are getting more out of AI than AI companies are getting out of us.

View the full article