“Generative AI helped workers avoid awful ideas, but it also led to more average ideas.” Kian Gohar, CEO, GeoLab
Harvard Business Review recently published an article on a study conducted by GeoLab.The basis of the study: Multiple teams from four companies tasked with generating ideas to solve a business problem their company faced.
Half of the teams were to use an open source version of ChatGPT; the other half were not given use of AI as an option.
The owner of the problem graded the results - and there was no significant difference between the two.
The assumption leading into the test was that the groups using ChatGPT would generate more and better ideas. The assumption proved false.
There was one significant difference. The group utilizing ChatGPT had 21% more confidence in their problem-solving abilities. The confidence was misplaced.
The Real Problem
The ability for AI to problem solve is real. The real problem is in expecting too much of it and in asking too little of it.
Expectation without work equals little leverage. Acting on generative AI, as iron sharpens iron, results in insight and innovation.
The article offers several insights:
- Craft highly specific problem statements.
- Train the AI by providing as much data related to your problem as possible.
- Consolidate team suggestions generated apart from AI, and then ask AI to analyze them.
- The more you question generative AI, the better its answers.
The Other Real Problem
Over time, normal sets in. We drift to center.
It’s true for people; it’s true for organizations. It’s true for countries.
Initially, our innovation attracted people who wanted what was distinctive. As time goes on, people ask for what is more normal. In meeting those needs, we lose our edge.
We get seduced into approaching things normally.
Generative AI is still newish, yet our approach can become normalish - we don’t train ourselves to keep AI trained.
The article recognizes this, which is why it calls for us to craft statements and to challenge answers.
But before you do this, you need to confront yourself: How normal are you treating generative AI (and how normal is your team approaching it)?
Better: How as an IT department are we approaching our problems? As we did yesterday, or as we know we must as if everyday is a new day?
AI has the possibility to open up whole new explorations; sad are those who are landlocked by drifting to the mean.
The weird travel new worlds.
