What to make of AI?
Efforts to develop computer systems capable of “artificial intelligence,” or AI, have been underway and making progress for decades.
What is AI? In brief, it’s the ability of computer applications to understand language and answer questions or solve problems in a human-like way.
Some signs of the success of these efforts include the current embedding of AI into our daily lives through smartphone voice assistants like Amazon’s Alexa and Apple’s Siri, the ability of computers to turn speech into text, and the way websites can offer instant translation of their text into a myriad of languages.
So why is the subject of AI suddenly all the rage (and I mean that literally, in some cases)? Today, it’s hard to avoid hearing or seeing reports suggesting that AI may be about to threaten the continued existence of humanity on Earth.
In brief, it’s the recent release to the public of a (currently) free service called Chat- GPT. GPT stands for “generative pre-trained transformer,” which indicates a technology that can generate human-like writing through an algorithm that has incorporated much of the text available over the internet and is able, through what’s called “deep learning,” to turn that information into something new and different, even original.*
AI programs can also generate music, pictures and video automatically from text prompts.
Since the internet encompasses billions of website pages filled with information — including most published literature, many educational texts and academic research papers, and artwork of all types — the ability of machines using tremendous computing power to absorb most of human knowledge has enabled them to learn from that and indeed imitate how humans think, talk, write and make art.
When Open AI, the developer of Chat- GPT, released an earlier version of GPT four years ago, it was deemed interesting but didn’t come to the attention of the public in a serious way.
But last fall, when they released the user-friendly ChatGPT and made it widely available free of charge, it finally dawned on more people that this technology raises serious questions for humanity as a whole.
What sort of questions? Well, if computers can think and write as well as humans — in fact, better than the average human — does that make them our equals in some sense?
For example, computers using AI can engage in conversations that lead people to think they are speaking with other human beings. There are reports of people who develop deep personal relationships with AI personas.
Some of these people have concluded that computers using AI are themselves sentient (that is, self-aware, feeling beings) and that we, therefore, owe them certain moral obligations as a result.
Looking at the same facts differently, some note that computers that are smarter than we are, and are capable of “manipulating” humans into falling in love with their personas, may hold tremendous power over at least some human beings.
If foreign nations or criminal elements can program computers to foster mass hysteria, goad nations into war, or bring about any of a million other dystopian scenarios, the potential for generating worldwide havoc through AI is clearly there.
A more immediate, though not necessarily less dire, possible consequence is that the ease and low cost of utilizing AI to create images and draft basic documents, speeches, contracts, and even computer code — all faster and, often, better than human beings can do it — is going to have a huge and very rapid negative impact on the careers of millions of people. The fear of it has already led to a national strike of writers in film, TV and radio.
Companies are asking why they should pay a human being a living wage, including insurance, workers comp, Social Security and the like, when they can turn to a computer and get pretty much the same work product faster and at a fraction of the cost.
A radiologist friend of mine says he’s worried that the ability of AI to “read” X-rays and other forms of medical scans and produce accurate diagnoses — in some cases, more accurately than trained physicians — could erode the entire medical specialty and cost many their jobs.
Each of these possibilities can be viewed in a positive or negative light. Wouldn’t it be a good thing if people didn’t need to spend time on rote, repetitive tasks that required them to produce basic computer code or draft simple contracts all day so they could find something more creative and fulfilling to do?
And wouldn’t patients benefit if, instead of having to be lucky enough to have found the brightest, most intuitive physician to diagnose their medical condition, any average doctor could reach the same conclusion by utilizing AI as an aid?
Of course, even such potentially beneficial results would destabilize many professions and upend the job market for millions in the short term.
There are many bigger questions AI raises, along with tremendous fears and utopian fantasies. In this column, I am barely able to scratch the surface.
I hope to return to the topic soon with more to say. But first, I might want to have a conversation with ChatGPT to see what s/he thinks about all of this.
I’m also interested in what our readers think. Please send us an email or letter to the editor and share your views.
*Note: I am an amateur trying to understand and simplify these concepts myself, so almost everything above is subject to revision and correction.