This column would have fit in well with the contents of the previous issue of SciTech†. But its subject matter presented itself more clearly in my consciousness later.
The various tech companies that are active on the AI scene are racing each other to create a more sophisticated AI agent — think: personal assistant. Something that goes way beyond “Alexa, what’s the weather forecast for tomorrow?” The rich and powerful already employ people — personal assistants — who perform all the mundane, but important and far from trivial, chores that we all would like to avoid. Things that can take an inordinate amount of time, but that by themselves do not really enhance our life experience. Such as: please file my taxes; make a medical appointment, taking into account its urgency and other items in my calendar; find and book an exciting two-week vacation for me next summer; see what’s in the fridge and give me a recipe for dinner that won’t take more than 20 minutes to prepare. You get the gist. It would be very attractive to many of us who cannot afford to employ such a person, to have such an (artificially) intelligent assistant at our disposal.
The concept of an AI personal assistant is fraught with numerous landmines
But it is also a concept that is fraught with numerous landmines. To begin with, envision the amount of very personal information this software will need to be entrusted with if it is to perform optimally. All those data will end up in the cloud managed by some company. People already provide a lot of personal information on the internet (whether they are comfortable with that or not), so they might give in; but whether that’s a good thing is, let’s say, debatable. Then there is the question of whom the assistant is really serving: will it really get you the best deal on your vacation, or might the tech company that provided the software be getting kickbacks from certain tour operators? We may also wonder what its effect may be on humanity’s skill set. People my age (and even younger) are well aware that most in the youngest generations have no clue anymore how to read a map and travel with the help of a sheet of paper. It is a skill that is being lost. Searching the internet (I am avoiding the use of a trademark word here) for information has replaced library research skills. The irony is that digital assistants may make us lose our internet searching skills. That means that we could also gradually lose the ability to check on the performance of the AI agents. In addition, we know that the information available on the internet is biased, since it is based on whatever digital data are accessible to feed the search engines. But becoming dependent on AI agents may lead to a further narrowing of the knowledge base — while losing the tools to hold the system accountable. That is bad for the users of those assistants, but even worse for those who cannot afford them. They might end up relying on an internet with severely diminished functionality. Thus, in turn, could further increase the gap between the haves and the have-nots in ways that are easy to imagine.
The moral will be obvious: this development needs to be watched closely and regulated carefully.
