Generative AI burst into public awareness with the advent of ChatGPT, which rapidly demonstrated its utility in various fields, including the legal industry. According to Above the Law, generative AI technologies, while not yet at a stage where they can replace significant portions of legal professionals’ work, have proven beneficial in streamlining tasks such as summarizing documents and conducting limited research (Above the Law). With ongoing improvements and the implementation of guardrails, it has become a credible tool for legal practitioners (Above the Law).
However, a new report from Goldman Sachs suggests that the progression of generative AI may face significant hurdles. The bank’s analysis indicates that the high costs associated with developing more advanced AI models may not be justified by the limited incremental performance gains that they’re likely to achieve. The report contends that the tasks generative AI can perform now are potentially the pinnacle of its capabilities, at least to a degree that rivals human performance (Goldman Sachs Report).
The looming question is whether the financial and environmental costs—such as the considerable electricity and water usage for cooling—are sustainable. Studies have flagged these concerns, revealing the heavy resource consumption required to train these AI models. According to Ed Zitron’s discussion on the Goldman report, improving the existing AI models will necessitate exponentially more data, which is prohibitively expensive (Ed Zitron’s Analysis).
For the legal industry, this may not be entirely negative. The current capabilities of generative AI already add value to legal work by accelerating mechanistic tasks. Efforts to make AI more “human” may be less fruitful than leveraging its strengths in productivity enhancements (Above the Law). Even if further advancements stagnate, generative AI in its present form could continue to be a valuable, albeit limited, tool for legal professionals.
In sum, while generative AI might not live up to its most ambitious promises, it is already contributing significantly to the legal industry. The pertinent issue now is whether the push for further advancements will yield enough benefit to justify the costs—or whether it is prudent to focus on perfecting the utility of what AI already offers (Above the Law).