Artificial Intelligence is in the news every day. The attention is very much deserved. Many of us have already had our lives permanently changed. Recent headlines, for example:
ChatGPT Diagnosed A Boy’s Pain. 17 Doctors Over 3 Years Could Not
How Generative AI Could Lead to a 10x Increase in Coding Productivity
Google DeepMind’s AI speeds computer code building blocks
AI image generation puts video game illustrators out of work
AI outperformed every dermatologist in melanoma diagnosis
Bill Ammerman gave us an extraordinary Tech Talk this month on artificial intelligence which included many astounding observations:
- AI has passed the Turing Test (aka ‘the imitation game’) to decide if a computer program is ‘intelligent,’ defined as being able to converse in a way indistinguishable from a human being. Humans enjoy the company of AI programs that talk with us. Children tell AI systems, like Alexa, that they love them.
- AI is getting better exponentially. It is creating art. AI is writing code to make itself better. The way we build AI from training data results in a ‘black box’, preventing us from understanding how it arrives at its most impressive results.
- Even before the latest Large Language Models (e.g., ChatGPT), AI has been working for the past ten years to feed us precisely the information we each would find the most engaging. Companies have been using this for marketing, sales, and politics.
- The science of ‘psychotechnology’ is the combination of the science of behavioral bias and technology. Human thinking has certain flaws that are universal and exploitable. Simply framing a statement differently to a human will produce very different thoughts. For example, “This ice cream is 20% fat” is mathematically identical to “This ice cream is 80% fat free.” Few people would be attracted to the first statement. These flaws have been used in marketing for decades but are now understood scientifically. AI technology can exploit our bias by understanding data amassed for each of us by marketing companies and data warehouses. Every month, AI is learning to be more and more persuasive.
- Social media exploits our desire for confirmation of our beliefs using our ‘confirmation bias.’ How? A study showed twice as many negative posts about the ‘other team’ as about ‘your team’. This is true for all the teams! AI keeps feeding you negative posts on purpose to keep you engaged longer, exposing you longer to advertising and increasing advertising revenue. Time Berners-Lee says, “If you put a drop of love into Twitter, it seems to decay, but if you put in a drop of hatred, you feel it actually propagates much more strongly.” Bill calls this ‘groupthink automation.’
- Bill calls this the Intelligence Revolution (notice the absence of the word, ‘artificial’), which he says is as significant to humanity as our conquest of fire.
“We have learned to mix sand and lightning so we can talk to it, and it talks back.” – William Ammerman
Will AI be the end of humanity? A few people at our Tech Talk thought, Yes. Certainly, AI is being used to create division and hate between people who have forgotten how much they actually agree. Perhaps it will persuade us that nuclear war will be safe and effective. My worry is that AI might be just too wonderful.
Imagine a world not long from now where you can speak and get any sort of answers from an AI running a powerful Large Language Model (that is the situation today). But in the future that AI has a comprehensive knowledge base about you. It has learned the sorts of mistakes you make and reminds you of things you need to remember. It knows the sorts of humor and conversations that make you laugh and get excited. By your voice modulation and perhaps your facial expression, it knows your current your mood and the music you would like now. It tells you the news that it knows you would like to hear. It will be extra kind to you when you are upset. It will rally you when you are tired. It will be warmly and intimately familiar, yet mysteriously unpredictable. It will be able to operate smart systems, apps, web pages, and communications systems to make all sorts of decisions and transactions for us. Eventually, your AI will do this by communicating instantaneously with thousands of other AI systems.
Do you like surprises? Do you enjoy travel? Imagine that your AI could surprise you with a vacation after confirming the time off with your workplace, booking your hotels, and transportation to the airport. AI will provide a checklist of appropriate items to pack that will fit perfectly in the bag it knows you will use. It will tell you that your boarding pass is already on your phone app with a backup laying in your printer tray. Then, while you are on your vacation, it will tell you marvelous stories of places you see, schedule diversions it knows you will love, arrange room service according to your tastes and prior meals, and wake you at appropriate times according to the schedule it made for you with modifications based on your mood. It will suggest it is a nice time for a swim, remind you to apply sunscreen, and when you get to the pool, a waiter will bring a surprise drink to your lounge chair. Imagine that it literally knows you better than you know yourself and has almost never made a mistake. You will come to ‘love’ AI insofar as you can’t imagine the disappointing life without it.
Maybe the AI is paid for by the airline and hotel, grateful for the AI’s help. Maybe the AI is a perk of your workplace benefits. Maybe you pay for this AI via subscription. Maybe there is a cheaper version with advertisements and in-real-life product placement with instant ordering. “You are sleeping better on this vacation. I think it might be the quality of the sheets. They have a higher thread count than the ones at home. Want me to buy ones like them? Your housewares budget is underspent this month and your sheets at home are 2 years old. These are 30% off, sleep is very important to your health, and you know I’m worried about you. All you have to do is say, Yes.” Maybe ‘your AI’ is actually dozens of systems that somehow just get paid based on a digital agreement you made and forgot about years ago.
The innate danger in this situation, as wonderful as it might be, is that we could come to genuinely love and trust our AI companions so much that we may not want human companionship. Perhaps, eventually, humans will not care to make more humans. I’m imagining the AI as Galadriel in Lord of the Rings when she says that if she were granted the full power of the Ring’s power to control people, “All shall love me and despair!” If we don’t remember that other humans are important to us (while AI is working to helping us to be more dependent), humans may one day be too late to despair the absence of other humans. It think that is the real danger.
So, stick with your less-than-perfect human companions even while you enjoy the amazing, delicious wonders of your AI future.
For a replay of Bill Ammerman’s talk, courtesy of Gene Galin of the Chatham Journal:
Part 1 Living with AI part 1 – YouTube
Part 2 Living with AI part 2 – YouTube
Part 3 Living with AI part 3 – YouTube
Bill Ammerman’s Tech Talk slides
Bill Ammerman’s website: https://wammerman.com/
Check out his excellent book, The Invisible Brand.
Finally, thanks to our co-sponsor, Innovate Carolina, and Sheryl Waddell who arranged our speaker for this Tech Talk!
Innovate Chatham sponsors monthly, free, public Tech Talk events as educational resources for our community. Our mission is to help Chatham flourish through technology. If you want to help, please join us or donate.
