LLM Links, 1/16
AI's get a tryout as teachers; Robots not ready for prime time?; creative uses of LLMs; Maharshi-Pandya on prompting LLMs to think slowly and carefully
One school in Arizona is trying out a new educational model built around AI and a two-hour school day. When Arizona’s Unbound Academy opens, the only teachers will be artificial intelligence algorithms
…The Unbound Academy's unconventional approach to teaching needed approval from the Arizona State Board for Charter Schools, which it received in a contentious 4-3 vote. Students in fourth through eighth grade will be enrolled in the program, in which academic lessons for two hours a day will be delivered by personalized AI, which will rely on platforms including IXL and Khan Academy. The idea pitched by Unbound is that it will make students happier and smarter, with more time to explore life skills and passions.
During those two hours, the students will be going through adaptive learning programs. While they study science, math, or literature, the AI will track their progress in real time. Depending on their performance, the AI will then adapt the curriculum's style and difficulty to help them succeed.
Pointer from Rowan Cheung. The typical result of an educational experiment like this is that it works as an experiment but fails to replicate outside of the experimental setting. This is sometimes called the Hawthorne Effect. The theory is that participants who know that they are in an experiment are often motivated to make it work.
“Some things which are very easy for people are very hard for robots,” said David Pinn, chief executive of Brain Corp, which provides software for automated floor-cleaning and inventory management robots used at retailers like Sam’s Club.
Even something as simple as picking up an arbitrary object and moving it “is a really hard problem in the world of robotics,” he said.
Until this is not such a hard problem for robotics, I am going to bet against anyone predicting AGI (artficial general intelligence) will arrive soon.
I often lean on ChatGPT to help with my kids. From resisting homework to boredom and frustration, the chatbot is a wizard when it comes to problem solving. It gives me some much needed clarity and often suggests ideas I never would have thought of otherwise.
Try, “My 8-year-old is refusing to do their homework. Can you suggest some positive reinforcement strategies and creative ways to make homework time more engaging?"
There are more ideas for using chatbots at the link. Many people do not appreciate that Large Language Models are creative. When you are doing Serious Work, their “creativity” (hallucinations) is annoying. But if you can get over your obsession with treating an LLM the way you would Google, you can get much more out of it.
Pointer from Mark McNeilly.
Maharshi-Pandya offers a set of instruction to prompt LLMs, including
Never skip the extensive contemplation phase. Show all work and thinking. Embrace uncertainty and revision. Use natural, conversational internal monologue. Don't force conclusions. Persist through multiple attempts. Break down complex thoughts. Revise freely and feel free to backtrack
Pointer from Zvi Mowshowitz.
substacks referenced above: @
@
If the Arizona experiment produces the same educational outcomes while only wasting two hours of the kids time a day, that would be a huge success. In fact it would mean productivity x4 and happier kids.
We are touring a school tomorrow that basically does three hours in the morning in a somewhat similar manner to this (there are teachers available) and then the afternoon is free project time.
There are other differences but the fundamental issue is there is so much waste in the traditional school setting. My kids don't need to sit at their desks learning nonsense while the teacher asks the boys to stay still for eight hours a day.
Arnold - You might be right about the Hawthorne Effect, but shouldn’t a business experiment be considered a success or failure based on profit and loss outcomes, or perhaps customer satisfaction and demand? Walmart and Standard Oil didn’t rely on the Hawthorne Effect when conducting their business experiments. I suggest we reduce the amount of scientific scrutiny in schools and instead treat them a bit more like churches or for-profit businesses. I’m glad to see such experiments taking place. Let’s hope this continues.