Until recently, the conventional wisdom was that while AI was better than humans at data-driven decision making tasks, it was still inferior to humans for cognitive and creative ones. But in the past two years language-based AI has advanced by leaps and bounds, changing common notions of what this technology can do.
The most visible advances have been in what’s called “natural language processing” (NLP), the branch of AI focused on how computers can process language like humans do. It has been used to write an article for The Guardian, and AI-authored blog posts have gone viral — feats that weren’t possible a few years ago. AI even excels at cognitive tasks like programming where it is able to generate programs for simple video games from human instructions.
While these stunts may be interesting, are they really indicative of what this tech can do for businesses?
What NLP can do
The best known natural language processing tool is GPT-3, from OpenAI, which uses AI and statistics to predict the next word in a sentence based on the preceding words. NLP practitioners call tools like this “language models,” and they can be used for simple analytics tasks, such as classifying documents and analyzing the sentiment in blocks of text, as well as more advanced tasks, such as answering questions and summarizing reports. Language models are already reshaping traditional text analytics, but GPT-3 was an especially pivotal language model because, at 10x larger than any previous model upon release, it was the first large language model, which enabled it to perform even more advanced tasks like programming and solving high school–level math problems. The latest version, called InstructGPT, has been fine-tuned by humans to generate responses that are much better aligned with human values and user intentions, and Google’s latest model shows further impressive breakthroughs on language and reasoning.
For businesses, the three areas where GPT-3 has appeared most promising are writing, coding, and discipline-specific reasoning. OpenAI, the Microsoft-funded creator of GPT-3, has developed a GPT-3-based language model intended to act as an assistant for programmers by generating code from natural language input. This tool, Codex, is already powering products like Copilot for Microsoft’s subsidiary GitHub and is capable of creating a basic video game simply by typing instructions. This transformative capability was already expected to change the nature of how programmers do their jobs, but models continue to improve — the latest from Google’s DeepMind AI lab, for example, demonstrates the critical thinking and logic skills necessary to outperform most humans in programming competitions.
Models like GPT-3 are considered to be foundation models — an emerging AI research area — which also work for other types of data such as images and video. Foundation models can even be trained on multiple forms of data at the same time, like OpenAI’s DALL·E 2, which is trained on language and images to generate high-resolution renderings of imaginary scenes or objects simply from text prompts. Due to their potential to transform the nature of cognitive work, economists expect that foundation models may affect every part of the economy and could lead to increases in economic growth similar to the industrial revolution.
A Language-Based AI Research Assistant
In my own work, I’ve been looking at how GPT-3-based tools can assist researchers in the research process. I am currently working with Ought, a San Francisco company developing an open-ended reasoning tool (called Elicit) that is intended to help researchers answer questions in minutes or hours instead of weeks or months. Elicit is designed for a growing number of specific tasks relevant to research, like summarization, data labeling, rephrasing, brainstorming, and literature reviews.
I’ve found — not surprisingly — that Elicit works better for some tasks than others. Tasks like data labeling and summarization are still rough around the edges, with noisy results and spotty accuracy, but research from Ought and research from OpenAI shows promise for the future.
For example, the rephrase task is useful for writing, but the lack of integration with word processing apps renders it impractical for now. Brainstorming tasks are great for generating ideas or identifying overlooked topics, and despite the noisy results and barriers to adoption, they are currently valuable for a variety of situations. Yet, of all the tasks Elicit offers, I find the literature review the most useful. Because Elicit is an AI research assistant, this is sort of its bread-and-butter, and when I need to start digging into a new research topic, it has become my go-to resource.
All of this is changing how I work. I spend much less time trying to find existing content relevant to my research questions because its results are more applicable than other, more traditional interfaces for academic search like Google Scholar. I am also beginning to integrate brainstorming tasks into my work as well, and my experience with these tools has inspired my latest research, which seeks to utilize foundation models for supporting strategic planning.
https://hbr.org/2022/04/the-power-of-natural-language-processing