A.I. for Leaders: the T.I.N.K.E.R. Framework

A.I. is everything, everywhere, all at once these days. Believe it or not, Chat G.P.T. — the large language model (L.L.M.) released by Open A.I. that started this frenzy — won’t be a year old until?November,?yet it sometimes feels like we’re not ready for what’s already happened.

Over 100 million have registered to use it in its first two months (the fastest adoption of a new technology in history), and Chat G.P.T. now shares the stage with strange-sounding tools like Bard, Claude, Midjourney, and Stable Diffusion.

Many enterprising individuals have already started incorporating Generative A.I. chatbots into their workflow to great effect. In some ways, it’s easy to know what to do if you’re a knowledge worker or a company of one: just start playing with these technologies and see what they can help you do.

But what should?leaders?do right now?

First,?let me reassure you: it’s not too late to seize this opportunity. As one pundit said,?we are in the AOL “you’ve got mail” stage of Chat G.P.T.?But you must incorporate G.P.T. 4 (the latest public iteration) into your thinking?now?to have a strategy for G.P.T. 5 and be ready for the real game-changer — G.P.T. 10 and beyond.

Here is a framework to structure your first steps in this new space. It’s designed as a step-by-step approach to familiarize yourself with the promise and perils of these tools, encourage your teams to apply them, and place sensible guardrails around their use and wide-scale deployment. Ultimately, a key goal of the T.I.N.K.E.R. process — suggested by the acronym itself — is that you use it to develop an AI-friendly posture. By incorporating this model into your leadership approach, you are setting important intellectual and organizational habits in place.

Try it yourself:

It’s only by interacting with these tools that you can get a real sense of their potential, as well as the limits of their current abilities. Equally important, you should lead by example. Executives can’t expect their team to adopt a behavior unless their boss models it first. So try simple prompts at first, and here are simple but powerful hacks. To guide it best, start any query with a role you’d like it to play and use “act as if …” as your opening line, then get as specific as possible in your instructions. Pro tip: use as many adjectives as possible to describe the situation.

However, incorporating it into your workflow won’t trigger team-wide change on its own; you will also need to …

Incentivize your team to use it:

First, provide resources to speed adoption: one C.E.O. is paying $2,000 a month for his entire company to access Chat GPT+ (the latest and most advanced version of the tool) and estimates that it’s already paid for itself. Second, could you use your spotlight to call attention to team members who develop best practices and great use cases? You can encourage lunch-and-learn sessions to share them and regularly recognize the A.I. vanguards in your organization. Validation is not enough, however. Make skillful A.I. deployment a part of your employees’ next performance review, but make sure the applications of these tools are more than just performative.

This is why you must also …

Number-crunch:

Track the R.O.I. of your early A.I. experiments. Evaluate their effectiveness by measuring their impact on automating mundane tasks, accelerating processes, as well as corresponding time and cost savings. Used skillfully, these tools can supercharge your organization's output.

You can reassure your team that incorporating these technologies will not replace the role of human judgment in your processes because a master A.I. skill today is to …

Know how NOT to use it:

Each A.I. tool has specific Achilles heels. Many don’t handle math problems well. Most produce answers in robotic, bland prose, so you must add your own “special sauce” to give any written content some spice. All L.L.M.s are prone to what the experts call “hallucinations” — the term of art for when A.I. makes stuff up, so you have to sense-check everything it spits out. At this point, you need to keep a “human in the loop” to review and verify the outputs.

It’s essential to avoid sharing proprietary information with these tools, as it may become public. Since these A.I. models send user data back to developers for cloud processing and future improvements, anonymize any data you enter into them. This is why Apple has restricted its employees’ use of?ChatGPT?and A.I. coding tools such as GitHub Copilot.

But the only way to learn all of its capabilities?and?limitations is through experience, which is why it’s crucial to …

Explore constantly:

Be on the lookout for new use cases, updated versions of the bots you already use, and of course, new tools. A.I. can be used to automate activities, but it can also be used to develop breakthrough products and services. One Tech C.E.O. I know started an entirely new business leveraging the power of Chat G.P.T. So as a leader, it’s crucial to foster a culture of exploration and experimentation across your organization where you?and?your team stay continuously curious about these technologies.

But beware of over-using these tools, and please …

Resist being de-skilled:

I’ve always said that it’s easier to edit than it is to write. L.L.M.s like Chat G.P.T. and Bard make getting a first draft in seconds frictionless and effortless. Wonderful, right? Yes, but it’s often in those frustrating early attempts that we learn a lot — including what?not?to do or add to a product or service. Relying on these tools to get us started risks atrophying some valuable abilities. I call it the?Google Maps Trap: just as your knowledge of a city suffers if you rely on G.P.S. to get around, so will your creative skills if you subcontract all of that essential initial work to a bot.

Will AI Augment or Automate You?

That’s up to you. T.I.N.K.E.R. with it, and you will develop an AI-embracing culture within your company. You want people who are creative in applying this new technology and have an experimental and entrepreneurial mindset toward future tools. You want to build a team of Tinkerers.

That culture and posture will be crucial for the day — coming sooner than we realize — when these tools are ubiquitous, and their impacts are tumultuous. Chat G.P.T. and the other L.L.M.s have already altered everything from individual workflows to public company valuations. It will doubtlessly disrupt entire industries and transform society.

If you’re a leader, know this: there is no way to “run out the clock” on this revolution or pretend it won’t affect you or your business. It’s already here, and its effects are accelerating. The only path forward is to?lean into this technology,?learn its strengths and weaknesses, and?leverage its power?to improve your organization’s performance.

In other words,?executives must adopt a practice of I.A. — intelligent augmentation — to meet the challenge of A.I.