In 2023, we witnessed an explosive proliferation of Large Language Models (LLMs), including ChatGPT, which made a profound impact on various industries. In less than a year, ChatGPT has garnered an astounding 100 million weekly users, while over 2 million developers have harnessed the ChatGPT API to craft innovative solutions across numerous Fortune 500 companies.
In the coming year, we anticipate a shift in focus towards the development and adoption of smaller AI models specifically tailored for vertical applications. These models are designed to address the unique requirements of specific industries and domains.
Typically, data used for training in these vertical applications is more homogenous and of higher quality compared to the broad data spectrum used in training larger LLMs like ChatGPT. Consequently, solving specific tasks with generative AI may not necessitate the high computational costs associated with LLMs aiming to be a universal assistant, catering to a wide array of professional needs.
Vertical application use cases are expected to fall into two primary categories:
First Draft Generation: One such vertical application involves the development of software tailored for legal professionals. Notably, Thomson Reuters, the creators of Westlaw, recently acquired Casetext for a substantial $650 million investment. Casetext boasts an AI legal assistant tool that claims to extract information from contracts within minutes. Thomson Reuters is already in the process of unveiling a suite of AI-based products, including case law research and summarization. While there have been instances of attorneys using ChatGPT to discover that the generated case law was fictitious, vertical applications like legal software offer a more promising solution. Thomson Reuters, with its extensive library of case data, can cross-verify the output of its generative AI model, substantially reducing the risk of hallucination. Therefore, specialized legal software, trained on specific data types, holds the potential to generate high-quality first drafts. Several startups are already working on automating the initial drafting of patent applications, showcasing the growing potential of AI in the legal field.
Company/Field-Specific AI Assistants: In addition to first draft generation, vertical applications may focus on developing company or field-specific AI assistants. With OpenAI’s release of GPTs capable of creating custom versions of ChatGPT for specific tasks, the potential for tailored AI solutions is more accessible than ever before. Many law firms maintain vast repositories of employee training data and examples of task execution. An AI assistant can prove invaluable in onboarding new staff and swiftly providing answers or creating templates for various tasks, streamlining operations and enhancing productivity.
Ultimately, the long-term adoption of these AI solutions hinges on a critical cost-benefit analysis. For a paradigm shift to displace conventional practices in any vertical application, the AI output must deliver significant time savings and cost advantages when compared to human labor.
As we venture into the future, the trend is likely leaning towards the development of specialized, cost-effective, and accurate vertical AI applications that cater to specific industry needs. While challenges remain, the potential for AI to revolutionize productivity and efficiency in these domains is undeniably promising, heralding a new era of innovation and automation.
Become a subscriber of App Developer Magazine for just $5.99 a month and take advantage of all these perks.
Discussion about this post