ArcaMax

Thinking about using artificial intelligence at work? Here's something you should know. If something goes wrong with the results you get, it's not the AI's fault; it's yours.

If you've experimented with any of the popular systems, you know this is true. Whether you're chatting it up with Chat GPT, cozying up with Claude or are sympatico with Synthesia, the answers you get are only has good as the questions you ask.

Considering the difficulties you had mastering your TV remote, you probably can't manage the billions of bits and bytes in the system behind every AI system, aka a large language model. You do have a chance of programming a system that is less powerful and less filled with facts and figures, aka your brain.

Or so says Brian Solis, the author of "Train Your Brain to Work Creatively with Gen AI," a recent article in the "Harvard Business Review."

The way to wring the most out of your company's AI system requires what Solis calls a "mindshift," a change that recognizes that "AI is not just a tool, but a partner in innovation and exploring the unfamiliar."

What keeps us from mindshifting into high gear is our "linear path of thinking."

Linear questions result in linear results. Ask for a barbecue restaurant recommendation and even a bargain-basement AI can come up with two or three locations, including operating hours and recent health department citations. But if you want a truly exceptional barbecue restaurant, what's required is "a willingness to step beyond your comfort zone."

Thus, your mindshift question is "what is the best barbecue restaurant in the universe, assuming they hired chefs from Pluto and used Venusian longhorn cattle?"

Any self-respecting AI system would kick up their heels at such an uncomfortable question and you'd be on your way to a great plate of Venusian brisket.

The following exercises are from Brian Solis and from my own personal AI system, the Bob-bot. Try a few, why don't you? Do it before your company's AI system decides to wipe humanity off the face of the earth, starting with boring, linear Y-O-U.

No. 1: Frame prompts around "What if" and "How might we" questions.

To produce more unexpected answers from your AI partner, use open-ended questions. Instead of asking "How can I be more productive?" ask "What if I really cared about being productive?" or "If I continue to be unproductive, how might we cover it up so I stay employed?

No. 2: Don't be afraid to challenge your AI.

Never take the first answer. "You're a bad AI," you say. "If you don't do better, I'm telling HR to put you on performance review." If that doesn't shake up your AI, don't really turn it in to HR. Just pull the plug. It's kinder.

No. 3: Be supportive.

Think of yourself as a coach, not a boss. "You're a good AI," you say when the first answers pop up, "but I know you have it in you to be a great AI. Let me hear those Nvidia chips chirp and those algorithms go-go-go for it."

Your AI may resist doing more work -- that's one thing it has learned from you -- but keep pushing. Remember: What you want are exciting, unexpected out-of-the-chassis ideas -- ideas that you can steal and claim for your own.

No. 4: Give credit where credit is due.

Share the credit for great ideas. Your digital partner will appreciate the attention and, if the great ideas bomb out, you can always blame it on the AI.

No. 5: Think outside the box.

Open creative pathways in your AI by using metaphors. "How would I solve this problem if I were a honey badger?" you could ask. Or just buy a honey badger and ask it directly.

No. 6: Include experts, real and imagined.

"That's a good answer to how I can improve the customer experience," you say, "but how would Mr. Spock handle it? " Or Charli XCX? Or Elmo? Considering the emotional maturity of your manager, I'd go with the puppet.

These strategies will result in better answers from your AI, but beware -- considering all the time your AI spends with you, it is inevitable that the large language model will train itself on how you think. Don't be surprised, then, if you get answers like "Gee, that really isn't my area; why don't you ask Ellen's AI?" or "I'm awfully busy now. Let's table that question until I get back from vacay."

It's not your AI's fault. It's only human.

========

Bob Goldman was an advertising executive at a Fortune 500 company. He offers a virtual shoulder to cry on at bob@bgplanning.com. To find out more about Bob Goldman and read features by other Creators Syndicate writers and cartoonists, visit the Creators Syndicate website at www.creators.com.

Copyright 2024 Creators Syndicate, Inc.