Are you a Centaur or a Cyborg?
A team of Harvard social scientists and BCG consultants published a fascinating working paper on Monday, demonstrating the effect of Chat GPT4 on knowledge worker productivity.
Across 18 different realistic tasks, BCG consultants using AI completed more tasks (+ 12%), more quickly (+25%), with higher quality (+40%) than those who were not using AI.
Not surprisingly, the consultants who score the lowest on baseline assessments for these tasks realized the most improvement with AI. But even the top consultants improved.
Those who benefited the most from the positives of GenAI, without getting fooled by hallucinations, used one of 2 approaches.
The first group drew clear lines of separation between AI and human tasks based on relative strengths. For example, they used human judgment to define the problem and determine the best statistical techniques to use, then turn to AI to run the code and produce the graphs. The authors called them Centaurs (clear line between human torso and horse body).
The second group integrated AI and human activities in a more intertwined way. For example, they write a prompt, suggest edits, and add details to build off AI’s initial version in an iterative fashion. The authors called the Cyborgs.
The titles are descriptive, despite the mixed metaphor of Greek mythology and 1980’s Arnold Schwarzenegger films.
Which are you?
#ArtificialIntelligence, #Productivity, #AnalyticsForImpact