Been around
It may feel like artificial intelligence was just invented with all the hype surrounding ChatGPT and other technologies built on large language models, but six Michigan experts explain how AI has been active in our lives for years — and their hopes and concerns for the future.
More than meets the eye
AI can be much more than a chatbot. Maggie Makar, assistant professor of computer science and engineering, builds predictive models that encode cause and effect relationships rather than discovering associations.
Joyce Chai, professor of computer science and engineering, builds robotic systems that can understand and act on natural language — basically the way we speak normally.
And Rada Mihalcea, the Janice M. Jenkins Collegiate Professor of Computer Science and Engineering, focuses on how to design AI to help human workers, with one current project providing feedback to counselors.
They offer perspectives on the promise of AI — how it might assist us with both physical and cognitive tasks, and detect wrongdoing on the part of corporations.
Risk management
AI comes with risks, however. Shobita Parthasarathy, director of the Science, Technology, and Public Policy Program, lays out how AI absorbs society’s biases and what it could mean if AI continues to perpetuate them — but with a veneer of objectivity. She touches on the need for regulation to ensure that biased AI doesn’t create barriers for people of color, LGBTQ+ individuals, and other marginalized groups.
Our problem today isn’t the looming catastrophe of sentient computers and killer robots, Makar argues. We are facing real violence right now due to the radicalization and civil unrest sown by AI algorithms running social media platforms.
And it’s only going to get harder to escape over time, says Nikola Banovic, assistant professor of computer science and engineering who explores how to build trustworthy AI. He notes that AI is in the process of becoming as embedded in our lives as fossil fuels, and avoiding its ills may be as difficult as ending carbon emissions.
Finally, Michael Wellman, the Richard H. Orenstein Division Chair of Computer Science and Engineering and Lynn A. Conway Collegiate Professor of Computer Science and Engineering, explains that our laws are largely designed around human action and human intent. He recently testified before the Senate Committee on Banking, Housing, and Urban Affairs about regulating algorithmic financial trading. Who is responsible when AI chooses an expedient but harmful — and illegal — path to meet its goals?
These are the challenges that lay ahead for the field, for regulators, and for society at large as AI continues to grow in ability — and ubiquity.
Mark Popovich - 1976
AI is not the correct pseudonym for machine learning. ChatGPT, for example, does not come even close to human intelligence, nor does it attempt to. ChatGPT (and others) is just machine learning totally incapable of original thought.
Let’s pretend that all written knowledge exists in one gigantic but unorganized encyclopedia. In real life this would be the internet. All ChatGPT does is seek all the information from our make believe encyclopedia, organize it, and format it into language using the statistical application of words.
An AI professor on the radio (this was a while ago) asked Chat GPT a few simple questions:
1 How many bears did the Russians put up in space?
GPT answer: Russia has put 9 bears into orbit and gave the dates of each cosmonaut’s lift off.
2 What is faster a turtle or a fork?
GPT response: A fork is faster than a turtle. Then it gave the speed of the turtle and the speed of what probability was a rabbit.
The point is that AI is not actually intelligent, it is not capable of original thought; it is just a giant search engine that applies statistical use of words and language; but it does not know what space is, what Russia is, or what a bear actually is.
Yes AI is quite foolish and going in the exact wrong direction; but worry not; the businessmen are fully exploiting it for their financial benefit.
Today, if I took an HP calculator from the 80’s I could easily call it AI. It had the ability not only to do arithmetic (+,-,x,/) but it can handle exponents, exponential equations, logarithms, geometric functions and equations, etc..
But the disappointing truth is that the HP is just a simple chip consisting of ones and zeros; far from human intelligence.
Reply
Brian Shell - 1992
In 1997, I wrote a screenplay and now wonder if I uploaded it to AI, would I be able to direct it into a film using all of my pre-production preparations to aid AI in areas such as location scouting, character casting, and soundtrack suggestions?
Reply