TL;DR — The AI industry is booming with startups focusing on uncovering riches via text summarization, chatbots, information retrieval, and content generation. But these areas are not where the real treasure is buried.
AI’s untapped potential lies in customized explanations. It is not about exploiting what we already know but about exploring and understanding the vast territories of what we don’t fully grasp.
In October 2022, there were about 50 generative AI startups, most of which had only launched a year prior. As of July 12, 2023, a mere 10 months later, that number had boomed to 335, according to CB Insights. Many of these newcomers have built their business models around text summarization, chatbots, content generation, and text analysis, such as sentiment analysis. Others are trying to figure out how to integrate LLMs using their data (indeed, many B2B companies are jumping in with solutions). Many businesses and organizations are even using AI to assist in writing code.
Yet as incredible as this new wave of AI tools is, it is just scratching the surface. The ability to distill a Tolstoy novel into a tweet or pen a sales email that sounds like Shakespeare on a coffee break is impressive but does not capture its full potential.
We have found many glittering fragments, but we are still digging in the wrong place.
One obstacle that might have made companies hesitant to fully explore the explanatory power of Large Language Models (LLMs) is their tendency to “hallucinate,” or produce information that’s not accurate or factual. This concern is valid but not unique to explanatory applications of AI. Even in other common uses, such as text summarization and sentiment analysis, LLMs might produce outputs that are inaccurate.
Further, technology that uses LLMs can now cite sources (Bing Chat, Perplexity, using LangChain, etc.), and new LLMs are being produced that reduce hallucination, so output accuracy is at 95%! That is certainly more accurate than most human explanations of any topic.
One remarkable capability of LLMs like ChatGPT, Claude, and BARD is their ability to explain subjects, topics, and texts in a specific, customized way that resonates with the individual. The simple prompt “Explain like I’m…” [in 8th grade, a Marketer, a Lawyer, a novice, etc.] is where the true value resides.
AI doesn’t just hand you the CliffsNotes version of a subject; it walks you through it and answers your questions. If you let it, it challenges your thinking and even engages you in meaningful conversation. Even with all of its “hallucinations,” it can be a personal tutor who not only knows the subject matter inside out but also knows how to present it in a way that connects with you personally.
Human comprehension of a subject is what drives innovation. This direction in AI isn’t about quick fixes or shortcuts. It’s about enriching the learning experience, enhancing decision-making, and providing insights that go beyond mere surface-level understanding. It’s the difference between knowing the plot of a novel and grasping its themes, characters, and underlying messages.
The most obvious area for this remarkable ability of LLMs is in schools and universities. Teachers can develop lesson plans that adapt to each student’s pace, learning style, and prior knowledge, thereby fostering true comprehension rather than cramming facts that are forgotten as soon as the Red Bull wears off. But this industry can take a while to catch up with the times (remember how long we couldn’t use the internet as a citation source?), so I will not dwell any longer on this subject. I will say, though, that internationally, the United States ranks 25th of 37 OECD countries for mathematics education, so perhaps it’s time we catch up.
If we were to focus our AI development efforts on explanation, here are some scenarios and applications that could translate to financial outcomes:
1. Corpotrate Training and Development
2. Employee Onboarding
3. Cross-functional Collaboration
4. Healthcare Decision Support
5. Legal, Compliance, and Regulation
6. Sustainable Development and Environmental Planning
7. Crising Management and Response
8. Ethical and Social Responsibility
9. Customer Education
These are just some of the areas where businesses can reap immediate rewards. The implications for the power of explanation are vast and can’t be covered in one article, or even a novel. I’ve focused on the scenarios above because financial gains are what tend to catch the most immediate attention and spark innovation.
The journey towards innovation is not merely about exploiting what we already know but exploring and understanding the vast territories of what we don’t fully grasp. We can enter new realms of innovation that will not only enhance our individual capabilities but elevate humanity. The true gold in the AI landscape lies in our willingness to dig deeper, seek comprehension, and innovate with wisdom and empathy. Let’s shift our focus from the gold rush and recognize where the real treasure lies.