Conversational and generative AI

These two letters, AI, have risen to top of tech talk and media minds.  A quick search reveals near hysteria in the media and near panic among leading tech firms.  And even the Federal Trade Commission is wary of the current hype and AI-powered marketing claims.  Experts warn about the need for ‘guardrails’ against tools giving inappropriate answers, and they worry about ethics to underpin the answers of tools like ChatGPT. So as a result, tech firms want you to know that they are working on appropriate fixes.  As is typical of hype cycles, initial innovation is viewed as astounding, followed by efforts to break or downplay it.  Which means it will become very useful.

Can tools like ChatGPT and DALL-E be helpful to older adults? Of course, in the ways it can be helpful to any adult. The simple examples – retrieval of information and generated images seem very different from traditional (ad-driven) search tools. Given the early user base of ChatGPT (optimizing language models for dialogue), 100 million within the first few months, clearly developers, investors, and consumers agree. And multiple other companies are jumping into the market, perhaps playing catchup with early entrants. DALL-E (creating images from text captions) and other generative AI have become, uh, visible, and perhaps really useful in film-making and video game development. But how will this category be applied in the older adult marketplace, either for them directly or for those who serve them? 

Occasionally some tech really is NEW! At the end of November, the tech world figuratively-speaking fell over when the staggeringly well-funded AI tool, ChatGPT, was launched by OpenAI, its sponsor and creator .Within five days, a million were giving the tool a try – and who knows how many by now (six weeks later, perhaps more powerful servers are required). Trying to understand the utility of an AI tool that knows so much about so much, some bloggers like Margaret Manning are very excited at the possibilities. And some of the ‘best’ examples are posted.  Consider that ChatGPT’s data sets were loaded as of 2021 – so the newest information isn’t there, at least for now. It also provides charming but incorrect answers to many questions and has been ‘banned’ temporarily from some sites.  Bah, humbug.