This is given as data: AI increases your performance. But … is that basically?
A bunch of scientists from Business of Harvard, Mit and Warwick universities with Wharton School in Pennsylvania wanted to discover. So they gathered a team of 758 employees of Boston Consulting Group and conducted them to several experiments to examine the impact of geni on their work.
Test
Investigators separated employees, all consultants, into two groups in various experiments. In each cases, participants were first asked to solve the problem without using artificial intelligence to determine the base value of performance. Then their group received a challenge in which some consultants could use GPT-4 while others didn’t.
In the first group, the task was to develop footwear ideas for area of interest markets.
For the second group, scientists deliberately designed a task in which AI will make a mistake in their answer. In this challenge, employees had to solve the problem for the company, assessing interviews from the outside and assessing quantitative data. Will consultants rigorously check the work of artificial intelligence?
Finally, the researchers measured how productive everyone was.
Results
In the first group (which had a brainstorming of footwear ideas), generative AI helped a lot. On average, consultants who use it NO have access. The quality of their work was also 40% higher. Interestingly, generative artificial intelligence turned out to be at the level of skill: it increased the performance of all-but less experienced consultants increased by 43% compared to those at the highest level (which improved only 17%).
In the second group dealing with the company’s problem, it is not surprising that the result was completely different: AI he worsened their performance. Consultants who used GPT-4 achieved correct answers only in 60% to 70% of the time. But those that went without – relying only on human brains – were right in about 84.5% of cases.
What we have learned
We already know that artificial intelligence helps us brainstorms and that it is not great in making critical judgments about complex information – which partly explains the results of those two very different tasks. But scientists gained further insight, watching people using GPT-4 in each.
The lesson is this: watch out. Although generative artificial intelligence significantly improves the performance of tasks in which it is good, it might throw you into things that are not. When consultants used artificial intelligence to perform more complex tasks, they were to blindly accept their results without a serious interrogation – the working version of falling asleep behind the wheel.
How to use it
“The key to Genai is to try it out for everything, a sense of what is good and bad,” says investigators Ethan Mollick, Professor Wharton and creator of the recent book Co -intelligence.
To use it more effectively, he suggests treating him like a human. “The best I know do not have engineering skills, but they are good at working with people,” he says. Even higher, tell artificial intelligence What Polite people to be. It has been designed to have a general default personality, but it might adapt as you would like. For example, ask for motion as a data maniac or a witty comedian. The first can be higher to analyze business questions; The second in writing ads that make people crack.
This is given as data: AI increases your performance. But … is that basically?
A bunch of scientists from Business of Harvard, Mit and Warwick universities with Wharton School in Pennsylvania wanted to discover. So they gathered a team of 758 employees of Boston Consulting Group and conducted them to several experiments to examine the impact of geni on their work.
The remainder of this text is blocked.
Join the entrepreneur+ Today for access.