By Recently I attended a post-legislative session “cracker barrel” where local legislators took audience questions and explained what had and hadn’t happened in the session. There was a lot to dissect, and while it was informative and provided insight into which bills were passed and which were killed and why folks voted the way they did, I found it frustrating.
Frustrating mostly because it wasn’t a conversation. A conversation allows the exploration of another person’s thought process and decision-making process. One never knows where a conversation will lead because it’s dialectical – it’s an exchange of ideas around a topic and that often leads in unexpected directions. A good conversation leads to greater understanding, to collective insights, to saying, “Ahhh, now I understand what you mean.”
This essay isn’t a conversation either, but I think we need to talk about artificial intelligence. A number of panelists admitted they didn’t understand how artificial intelligence works. Some said they used it a lot and believed that it would eventually replace all of their jobs, (others pointed out their jobs weren’t at risk because they were retired), and still others objected to the idea that their jobs could or would be replaced by AI. That said, they all concluded AI was here to stay and we’d just have to learn to live with it.
How AI works is pretty straight forward. Despite the implications of the name, artificial intelligence is just a massive data set which can be queried using an algorithm to provide a response to that query – it’s not intelligent; it’s not actually thinking, it’s responding based on how an algorithm directs it to handle the data set. Think of it as a best guess to a question or a task.
The AI application that seems to be proliferating most rapidly is the chatbot – it’s part of the promised AI revolution eliminating the need for people to respond to complaints. Unfortunately, customer service has gotten worse with AI as people fight with chatbots that don’t understand our problems (my kingdom for a human being!). A new study from Pew Research shows that public trust in AI continues to drop. The more people know about AI, the more they hate it.
Writing hasn’t improved; AI facial scans make mistakes and haven’t sped up airport lines and teenagers and adults who listen to AI “friends” have been harmed more than helped. AI slop is everywhere – this digital clutter saturates social media and other online platforms in the same way that spam floods email. It has become so pervasive we’ve coined a word to describe this general process. Enshittification describes the degradation in the quality and experience of online platforms over time.
Evidence abounds that the use of AI weakens the development of critical thinking skills in children and adults. On top of that, we aren’t yet paying what AI queries actually cost.
Currently, AI is bundled with almost all software, Microsoft, Apple, Google, etc. Part of the reason companies imbed it in their products is because if consumers had to pay for it separately many people would choose the cheaper AI-free version. But once they get us used to having it, the cost will no doubt rise exponentially.
Another issue concerns the data being collected and used to power AI. The bulk of that data was generated by human beings and much of it is copyrighted or ought to be private (like our social security information) and none of these AI companies are compensating the human beings who created the material. That legal issue, whether people are due compensation for their work or the ethical question of whether all our private data should be compiled in a giant searchable database, have yet to be resolved – enshittification. So we have three issues, horrible performance, high cost and numerous legal and ethical questions.
Supporters of AI claim performance will get better, and the legal questions will be resolved, but what about the prediction that it will replace everyone’s job? I know the panelists were talking about their “real” jobs not their legislative ones, but think about how much money we’d save if we no longer had legislators – shouldn’t we just replace legislators with AI? AI is clearly smarter – after all, it can answer any question, and it could write error-free bills and do it all in seconds (actually, I’m positive it’s already being used to write bills just like it’s being used to write everything else from legal briefs and books to newspaper articles).
Senator AI doesn’t require a building (no need to spend millions on repairing the Capitol) or for people to spend time on, as one panelist labeled it, the Smithfield West killing floor in Pierre. Lobbyists would have to figure out how to influence Senator AI when they can’t take the senator out for dinner and drinks or funnel money to its campaign. Of course, if you own the AI company, all of Senator AI’s decisions will be favorable to your position.
Again, according to everyone on the panel, AI is here to stay, and we have to use it. Two observations – the first is that AI (or should I say technology?) is really the only product I can think of where people believe we don’t have a choice of whether we use it or not. As I’ve mentioned, part of the issue is that companies force us to use it by not offering alternatives. But if everyone parrots the same line, we’ll be left behind if we don’t use it or, even worse, China will win the AI race and all of us will be at risk if that happens (though how not turning over decision-making to a machine puts us at risk from the Chinese isn’t exactly clear).
To summarize, it doesn’t matter if it doesn’t work well; doesn’t matter if it requires massive amounts of energy which raises energy costs for all of us; doesn’t matter if it uses acres of land or significant amounts of water or if it causes pollution or if it costs trillions of dollars or if it eliminates your job. We don’t have a choice. It’s progress!
The other comment we heard from the panelists is that AI is just a tool – it’s benign. It can be put to good use or bad use. Ignoring the fact that if the tool is a hammer every problem starts to look like a nail, the “it’s just a tool” argument shines a light on the biggest issue with AI. AI doesn’t know or care what a good use or a bad use might be – it doesn’t know or care what’s true or false or what’s right or wrong. If you tell AI to only tell you what’s true, it cannot do it. Only human beings can make those determinations, every time we simply accept what AI tells us we’re no longer making judgements.
I know AI can do lots of things very quickly that may not require ethical judgements – setting schedules and appointments, making lists, and organizing data and projects – things most competent managers and women with children do every day. But making good decisions requires knowing what’s true, and I’d hope none of us would argue decisions shouldn’t be ethical as well. To just say, yup, must be right because AI told me so doesn’t seem wise any more than accepting what some anonymous person on the Internet tells you about vaccines or what’s in your best interest or for whom to vote.
Case in point, one of the panelists opined that using AI was essential for national defense. Most recently an AI company, Anthropic, sued to reverse the Pentagon’s decision designating them as a supply chain risk over Anthropic’s refusal to allow the military unrestricted use of their AI technology. The Pentagon wanted unrestricted use to create autonomous weapons systems and surveillance systems – autonomous by definition means they operate without human intervention or oversight. They are automatic. This should scare the daylights out of any human being – especially when AI companies themselves are the ones warning us of the danger of doing this.
Consider the decisions in a war to target military installations vs. civilian targets like hospitals or schools, or to bomb boats and kill the occupants instead of boarding them and capturing those on board, those decisions require moral and ethical judgements – or at least they ought to. For example, when Hitler’s Final Solution for the “Jewish problem” came to light, the level of inhumanity horrified the world. People were stunned; how could this happen?
The Nazi perpetrators tried to defend their actions by saying they were just following orders, but ultimately it was determined they were responsible for their actions. Do you think “I was just doing what AI told me to do” will be seen any differently? Who will be held accountable if an autonomous AI weapons system targets schools or drops a nuclear bomb and kills millions? Will having no human to blame absolve us from the failure to do what’s right, moral and ethical?
That these questions aren’t being broadly discussed illustrates the need to explore the deeper issues surrounding AI rather than simply debate whether we should build hyperscale data centers because they’ll generate property tax revenue and a handful of “good” jobs and call it progress. History is full of examples of where billions of dollars are made producing and marketing harmful products – cigarettes for example. Once we discovered they caused cancer, we decided it’s unethical to allow companies to sell cigarettes without at least warning people of the danger of consuming them. In other words, just because we could do it, didn’t mean we should do it.
I’m not naïve. I understand there may be uses for AI, and that data centers vary in size and purpose, but when it comes to AI, I’m not willing to accept that it’s inevitable or that it represents progress. I’d rather pay for a person than a chatbot – it’s cheaper and at least that salary will circulate in the economy; it will support a person who will spend it to support other businesses that employ other people. It doesn’t erase a job. I’d argue as well that a person could understand and actually help me solve my issue. At the very least they could say “I hear you” and not simply repeat “our website can answer many of your questions – does that help?” Because no, it doesn’t.
Doug Allen is a retired educator and serves on the Watertown City Council.
