Generative AI Takes On Vertical Software – Fault Lines Appear
By Adam T. Hark
I began reading about it last year. In dribs and drabs, articles appeared noting how adept generative AI systems had become at producing workable scripts for software developers. The release (perhaps “unleashing” is more accurate) of OpenAI’s ChatGPT in November 2022 had already proven to be an inflection point in technological advancement. Many of its first commercial applications used in customer support and other front-line, expert machine (chatbot) use cases. But generative AI’s extraordinary ability to accelerate software development for companies while consuming a fraction of the traditional resources – time, money, and personnel – attracted the attention of professional coders and investors alike, contributing to speculation that at some point in the near future, the profession itself could be negatively impacted as businesses rethink their calculi when answering the “build, buy, or partner” question, relating to the development, implementation and deployment of new vertical software applications.
The first time I was exposed to this dynamic was while reading a March 2023 article in Business Insider, “Welcome to the final phase of software eating the world. SaaS companies beware, you’re on the menu this time.” The premise of the article being that ChatGPT had shown such a remarkable proficiency for outputting serviceable lines of code in response to developer queries, that it forced professional developers to question the consequences of its efficacy and efficiency on their future careers, and by extension, their products, in many cases, vertical software applications. In theory, companies would be able to reduce development staff and build their own applications on the cheap, and in a fraction of the time. Thus, the impact on the vertical software industry could be materially negative.
I read a few more articles that trickled out. “A Coder Considers the Waning Days of the Craft,” in the New Yorker, and another Business Insider piece that preceded the aforementioned story, “The end of coding as we know it.” All together, the reporting certainly provided credibility to the speculative concern. Though hardly a “hit the panic button” moment, to my mind it was surely something to keep an eye out for in 2024 as more and more companies begin to integrate generative AI into their businesses for the purpose of accelerating development and cutting costs, and, ultimately, deploying better software applications internally and externally. This dynamic would only grow, and for better or for worse, represented the beginnings of a fissure, a fault line if you will, between developers and software providers on one side, and corporate use of generative AI on the other, that would rip deeper and longer.
And then a second fault line appeared toward the end of 2023 involving generative AI more broadly, but, yet again, very relevant to the software industry. The fault line presented itself in the form of a lawsuit filed by the New York Times against OpenAI and Microsoft for copyright infringement. The dispute highlighted another growing challenge with the use and application of generative AI in its capacity to produce commercial work product based on LLM training data that includes proprietary intellectual property. The basic question being whether these generative AI systems can use anything they scrape from the internet to build their models. This same question would be applicable to the coding community and software firms as they might be unwittingly contributing their own IP to the LLM training data of these AI systems, which are in turn commercializing it in a way that’s disruptive and detrimental to themselves.
For additional color on this new dynamic, check out this podcast from last week: All-in with Chamath, Jason, Sacks & Friedberg, Episode 160: 2024 Predictions! Markets, tech, politics, and more… One of the podcasters actually predicts a downturn in funding and valuations in 2024 for the vertical SaaS industry as a result of generative AI.