ChatGPT’s promise is unfathomably deep, despite its habit of generating false information. Despite such shortcomings, many people are figuring out bona fide business applications for the technology.
In November 2022, AI startup vendor OpenAI released ChatGPT, and the AI world would never be the same.
Unlike other forms of AI that process large data sets in order to extract inferences based on those data, generative AI leverages such data sets in order to generate its own information, including text (like ChatGPT), as well as images, audio, video, and potentially other types of information.
OpenAI leveraged many terabytes of Internet-based information to train ChatGPT and continues to train it based upon the inputs of users around the world.
The result: ChatGPT can generate text that is eerily similar to human-generated text – similar enough to pass graduate-level tests, fool teachers, and frighten all sorts of writers who wonder whether AI will put them permanently out of business.
As millions of curious consumers, entrepreneurs, and others played with ChatGPT (as well as numerous other generative AI offerings springing up like weeds), the technology’s shortcomings soon became apparent.
Generative AI is often wrong. It generates incorrect, often misleading results. Furthermore, malicious inputs from humans can steer it to even more pernicious, biased results.
Given these limitations, there are serious concerns that generative AI is not appropriate for business use. The business world has no place for incorrect information, for fictions masquerading as the truth.
Fortunately, ChatGPT and its ilk are not the only games in AI town. Other generative AI companies are rapidly developing business-centric applications that avoid ChatGPT’s shortcomings to deliver unrivaled business value.
Creative vs. Factual AI Output
Despite ChatGPT’s shortcomings, many people are figuring out bona fide business applications for the technology. In large part, these apps provide creative output – in other words, output that resembles the creative output that humans produce.
By creative, we also mean fictional – generated artifacts that are not supposed to contain factual content. AI content from deepfake videos of celebrities to amusing images of elephants wearing hats to stories in the style of a particular author are examples of creative content.
There are business uses of creative content, but such content is intrinsically different from factual content. We want factual content to accurately represent the world – or at the very least, accurately represent the world as represented by the data we use to train the generative models.
Factual content from generative AI will typically have AI-generated elements that do not appear in the source data. An AI-generated summary of the news will be more than simply excerpts of news stories. It will ideally contain phrasing that better represents what readers want from a summary.
Even though these elements are in some sense creative, the overall output must be factual to have any business value. Nobody wants a news summary that mixes facts with AI-generated fiction.
Exploring the Business Uses of Generative AI: Analysis and Storytelling
Most business applications of AI focus on inferences from data – spotting defective widgets on a conveyor belt, identifying anomalies in operational IT data, and calling out the important words or numbers in a business document, for example.
Generative AI goes one step further, synthesizing inferences in order to tell stories about the data. For example, generative AI can create business reports, summarize information, and conduct research based upon massive quantities of source material.
Today’s generative AI can even analyze information, surfacing previously unknown connections among data and identifying common themes across different data sources.
Supporting Business Processes
Analyzing and telling stories about data are only part of the generative AI business story. The other side of the coin: optimizing business processes.
Generative AI can drive the optimization of information-driven workflows by changing the way that business leverage enterprise content management (ECM) platforms.
ECM is an extension of the established document management market that deals with document-centric workflows (and now other content workflows that might include emails, social media, or other types of content).
For many years, AI has played a role in ECM for the digitization of paper documents via optical character recognition (OCR). The state of the art of OCR today is vastly superior to the OCR from a decade ago, largely because of the role AI plays in the identification of characters, words, and figures in documents.
Generative AI takes the role AI plays in OCR one large step further, as it can extract, organize, and classify information, regardless of whether the ECM platform used OCR to digitize the source documents.
Generative AI can then analyze the information in documents to extract and summarize relevant information and make decisions about the appropriate actions to take based upon that analysis.
In other words, generative AI can automate content management processes, defining the process logic itself as an output from its analysis of the source data in the documents the ECM platform is processing.
The Intellyx Take
The explosion of ChatGPT onto the market has both focused attention on generative AI (and in truth, AI in general) as well as highlighting its weaknesses. Vendors of business-centric generative AI solutions are caught in this maelstrom, struggling to explain their essential differentiators in a sea of noise.
The smart business stakeholder will see through the turbulence to separate the factual AI solutions better suited for business from the creative AI landscape centered on consumer-oriented and fiction-generating applications.
Analyzing business data and telling stories based on those data is one important business use case for generative AI. Using such analyses to optimize content-centric processes is another. Only time will tell what other business applications of generative AI will arise in the coming years.
Copyright © Intellyx LLC. Charli AI is an Intellyx customer. Intellyx retains final editorial control of this article. No AI was used to write this article. Image credit: Craiyon.
