The far-reaching impact of ChatGPT-4 for complex tasks

12.2% complete more tasks

25.1% complete tasks faster

40% higher quality results

This is the finding of a study by Harvard Business School and Boston Consulting Group. For 18 different complex tasks, the researchers showed that consultants who use ChatGPT-4 outperform their colleagues who do not. 758 consultants participated in the study. The tasks that the researchers assessed involve typical work of a consultant that requires creativity, analytical ability and persuasive writing, among other things. For example, the consultants had to conduct a market analysis, draft a press release and convince colleagues of their product strategy with a memo. In doing so, the study proves the unprecedented power and transformative potential of language models, such as ChatGPT-4. This is in line with the feedback we receive from legal professionals who have participated in our trainings exclusively for legal practice. But there is more. The researchers also note that: – When performing a task inappropriate for GPT-4, consultants performed better (84%) without GPT-4 than with the use of GPT-4 (60%-70%). The explanation for this can be found in the fact that GPT-4 can also give a convincingly incorrect answer. This risk lurks when the user is not familiar with the (in)ability of a language model. – There is little to no understanding of all the capabilities and limitations of language models, such as GPT-4. The limit of GPT-4’s ability is invisible so it is unclear to users in which tasks GPT-4 excels and which tasks are challenging. ChatGPT can also give legal professionals an edge when performing complex tasks. However, Harvard research shows that a manual is necessary. This is so that ChatGPT can be used effectively and safely. With our training designed for and by legal professionals, we offer this. Want to know more? If so, please contact us. The research paper can be accessed at: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4573321