AI News Roundup: McKinsey economic report, AI-assisted human legal analysis study and more

Article co-written by Yuri Levin-Schwartz, Ph.D., a law clerk at MBHB.

There’s a lot happening in the world of AI. To help you stay on top of the latest news, we have compiled a roundup of the developments we are following.

McKinsey has released a report on the potential economic impact of generative AI. According to the report, use of these tools may contribute $2.6-$4.4 trillion annually to the global economy and half of all work may be automated by 2045.

The first controlled research study on AI-assisted human legal analysis demonstrates mixed results. A recent paper followed 60 law students from University of Minnesota law school who were given four legal tasks. The students undertook the tasks either with or without controlled access to GPT-4. Minor, inconsistent improvements in substantive work product were observed when using GPT-4. However, large increases in speed of task completion were shown consistently as well. The researchers believe that their results represent “important normative implications across the future of lawyering.”

The UK Judicial Office has sanctioned the use of large language models to write legal rulings. Published guidance warns the judges about the risks of generating misinformation and privacy concerns. Nonetheless, some judges seem enthusiastic about its potential. Lord Justice Birss of the Court of Appeal of England and Wales has referred to ChatGPT as a “jolly useful” tool.

OpenAI released a prompt engineering guide for improved interactions with GPT-4. The recommended prompting strategies included: 1) asking the model to adopt a persona; 2) providing examples; 3) specifying the steps required to complete a task; and 4) including important details to get more relevant answers.