Force Fitting Gen AI

Generative AI has been abuzz ever since OpenAI came out with ChatGPT. The craze is some much that every company, big or small, wants to somehow use it in their products, just for the kicks. The FOMO is so evident that several examples of companies using Gen AI seem force fitted and discordant. Lately, I came across at least two such use cases.
First, there is a Web3 or Distributed Ledger firm that came out with the ChatGPT plugin. The purpose — user could chat their way to get information on the balance in their wallet, buy coins, etc. — all of which could have been done by opening an app or a click of a button. If I log into my wallet app, I would expect to see my account balance on the screen without having to put extra effort to cajole a chatbot into giving me the data. All I need is the app to make a few API calls and retrieve the data from a database. The whole premise of this Gen AI use case sounded superfluous to me.
Another recent example that I came across was the announcement from a prominent cloud company that has an HR product in the market. They plan to integrate Gen AI with their job application system and use Gen AI to come up with job descriptions aligning with the role and the legal requirements of the country. On the surface, it sounds logical to use Gen AI. But there is a catch. First, there is no “describing” your requirement. No elaborate prompting and then fine tuning your prompts to eke out a plausible response from the bot. Instead, there is going to be a button, clicking on which, you will get a ready job description. Whether there would be a provision to refine the results, time will have to tell as it is not disclosed at the moment.
Now, the problem here is, this use case can be rendered without a Gen AI model. Imagine an HR screen to create a job opening. Typical fields I would have is a role, job code, job description, and location. In this case, I as an HR person would enter values for the role, job code, and location, leaving out the job description for the bot. The bot would be taking the inputs from the fields I entered and give me an apt job description. I could have a database where I store all the job codes, role titles, locations, and all the job descriptions used in the past for the role. If there are multiple job descriptions for the same role, I would write a round robin or a random number generator logic that recursively picks one of the options every time I click the button to generate a job description. I can still ensure the “generated” description would be relevant to the role and adheres to the legal requirements (as the past descriptions in my database would have aligned to the same regulations).
Both these scenarios where Generative AI feels redundant and excessive, are examples of how no one wants to miss the hype train, whether the destination is relevant to them or not.