Has it only been a year since we started to believe that artificial intelligence could actually save us from trouble and consign us to a life of idleness like George and Jane Jetson?
It’s only been a year since we started worrying about AI coming into our jobs, our spouses’ jobs, our kids’ jobs, and everyone else’s jobs except the CEO of OpenAI. Wait. Even that job wasn’t safe!
Those who oppose artificial intelligence look stupid
One of the biggest casualties of ChatGPT’s rise has been the arrogant attitude of hard-core, logic-loving professors who say AI will never come. One of my professors used to scoff at the term artificial intelligence. He liked to downplay talk about machines actually thinking. He liked to say that it would take decades, even centuries, for the intelligent machines in Star Trek to truly emerge. He sometimes said that artificial intelligence never stood a chance.
He’s lucky he retired before ChatGPT came along. Lucky because these new generative AI bots make it much harder for spoilsport logicians to say that computers can’t do much more than assemble NAND gates.
Electric bill worry
Considering the stocks of hydrocarbons we need to burn to power GPUs and TPUs with electricity, the planet might be a more sympathetic victim.
A real challenge for the world of AI is to find a way to unlock all the great opportunities without paying a high electricity bill. There is some hope that new chips, better algorithms and smarter use of layering in networks will save a few oil-filled supertankers. Will this be enough?
Attack on AI hardware
The big challenge for a new AI project is marshalling enough computing power to start learning. Demand is so high that GPU manufacturers like Nvidia can’t keep up.
Will this continue? While free markets have a way of eliminating scarcity, relentless growth and the big dreams of Silicon Valley can grow much faster than the market can deliver. There are also geopolitical problems.
Doomers and Boomers face off
The list of difficult-to-solve problems, such as Middle East politics, has become longer with the addition of discussions about what artificial intelligence will do to humanity. On one side are the Doomers, who see AI destroying jobs, social ties, and perhaps even all of humanity. On the other side, we have Boomers who are seeing lots of great gifts delivered to us while we relax on our virtual Lanai.
Who has the more accurate vision of the future? Experts and forecasters will debate this issue for months and perhaps years to come. If the answer were obvious, we would already know it. I used to joke that we should ask the AI, but companies have already advocated for them. It can be difficult to find direct answers about pressing issues.
The dangers of AI hallucinations
Is this artificial intelligence thinking? Or are you running a large statistical mechanism that chooses the next coin by rolling virtual dice? We know that the algorithm is just statistics, but is that enough to qualify as deep thought? What are the possibilities?
There are a number of metaphors that help explain what dominant algorithms do. Some like to call them “stochastic parrots.” Others like to think of them as a version of statistical compression algorithms like Huffman coding. We are still working to find the best way to explain the mix of genius and hallucination that results from these functions.
The dangers of AI accuracy
AIs tend to behave like children. Sometimes they make things up and that’s bad enough, but the real danger is when they start speaking the unfiltered truth. Some like the idea of truth-telling AIs and imagine they will bring more knowledge and understanding to the world. Others know that Jack Nicholson’s character in “A Few Good Men” was right when he said, “You can’t handle the truth” about humanity.
Lawyers at AI companies must be crazy having to defend anything that tells the truth. I asked Google’s Bard a disarming question about a living, breathing human being, in other words, the kind of topic that could lead to a libel suit. Bard told me in a rather sharp tone: “I am a great language model and can communicate and can answer a wide range of directions and questions.”