How Educators Can Easily Keep up with all the AI Developments

It’s not hard to see who is going to have jobs in the AI World. It’s not going to be students who are “scared anti-AI straight.”

Many educators I speak to express being overwhelmed with all of the advances in artificial intelligence and wonder how they can keep up to do what is right.

They are starting to see that AI is not a fad and that students who are left out of the AI world are going to be left behind.

I can’t tell them that it’s not hard to keep up with every single AI advance. Why? Because everyone agrees that, at this point, we are experiencing an exponential change in AI development and that exponential change is not something humans can absorb. It explains part of the reason that some people have chosen to ignore AI developments.

But it’s also easy to keep up and plan for where this is headed. Why? Because educators just need to start thinking about how to plan for a world where machine intelligence is “competitive” with human intelligence; a world where machines exceed the intelligence of every human in all or nearly all domains. That will almost certainly happen in the lifetimes of all or nearly all the people reading this, and it will certainly happen in all of our students’ lifetimes. Educators need to start preparing staff and students for this world. In some narrow areas, it’s already happened, and the change is rapid.

What are a few examples of rapid change?

Prompting. Until last week, ChatGPT4 allowed a prompt of nearly 100,000 words/125,000 tokens, and Claude allowed 150,000 words/200,000 tokens. Last week, Gemini said, OK, now you can enter a 700,000 word/1,000,000 token prompt. A few days later, Microsoft said they figured out how to allow a 2,000,000 token prompt. And as a side note, they already said Gemini can probably handle a 10 trillion token prompt (around 7.5 million words), and a new start-up, Magic, can already handle 3,000,000 words, at least for coding purposes.

Now, no one person is going to type out a 100,000+ word prompt, but what this means is that a person can upload a document with 1-2 million+ words and get questions answered about it in the prompt line. That’s pretty useful if you have a large dataset to work with. And where will we be next year?

Robotics. It’s hard to know if we are really in an exponential in robotics, but efforts here are massive. According to Business Insider,

Figure AI Inc. is raising about $675 million in funding with the help of investors like Jeff Bezos and the Microsoft Corporation, according to a Friday Bloomberg report, citing people familiar with the situation The artificial intelligence startup aims to create autonomous humanoid workers that can support labor shortages amid a waning workforce, the company's website says. The machine, called Figure 01, is an AI-powered, self-reliant robot with the ability to think, learn, and interact safely alongside humans, according to the company. The robots will be tasked with performing dangerous warehouse jobs for which there aren't enough workers.

This is feasible. Robots are advancing quickly, and a warehouse is a controlled environment that can be fully mapped using computer vision. Robots can learn quickly in such spaces.

Now if you have your “BS” detector on, you. may realize that there aren’t that many “dangerous” warehouse jobs and there aren’t that many unfulfilled warehouse jobs, so why do you think one robotics company got a $675 million investment to build robots to do the jobs of warehouse workers? Was it really to protect a low number of injured workers? Was this a charitable investment? These are good “higher order” thinking questions.

Will bots really be smarter than us?

Yes, they already are in some ways. While there are things they cannot do yet (engage in advanced reasoning, plan), they can translate 202 languages in all directions with less than 2 seconds of latency, they can analyze more data than any human could to make a decision, they can already diagnose tumors almost as accurately as doctors in many instances (and, as Geoff Hinton says, will soon exceed their abilities in this area), they have more general knowledge than any single human, and they are developing specialized knowledge. Many of the world’s most highly qualified AI scientists (Ray Kurzweil, Demis Hassabis) believe they will be at least as smart as us in all ways in which we are smart by 2030. If it’s 2040, does that matter? 2030 will be around the time this year’s entering freshman graduate. 2040 (also speculated by many as the time by which we will have quantum computing) is around the year today’s first graders will graduate from high school.

What about AI’s limitations?

Yes, AI has limitations, but “faster than humanely possible” research means these problems are being overcome.

Remember the old, "It can’t produce bibliographies” problem. This has been solved. Use Consensus (no hallucinations) or Perplexity.ai (very few). There are lots of other similar AI tools.

“But math, it can’t do math.” It still can’t do math perfectly (it’s a large language model…), but now it’s competitive at the Math Olympiad, and once models that can reason are fully developed, this won’t be an issue, as it will be able to reason through the math problem.

And, yes, it’s not yet a perfect customer service bot. I’m sure that many readers saw the story of an airline having to pay for bad instructions given by its bot.

But this is just one example, and substantial progress is being made in this area, with one new customer service bot company, Sierra, getting positive results.

Will these bots ever be perfect? Maybe yes, maybe no. But companies also know their employees aren’t perfect and that humans also make costly mistakes. And, even more obviously, they know the costs of mistakes made by bots may be far less than employing all the human customer service agents they do now.

Broadly speaking, what does this mean?

It means that AI continues its rapid advance toward a world where it can exceed human-level intelligence in at least many domains.

This means that more and more of our students and workers will have the opportunity to work with highly intelligent AIs. This is the future of work, commerce, and any online communication.

Can you give some specific examples?

Practically speaking, it means —

*Yes, AIs, when used properly, can write at least most of students’ papers as well or better than they can. AIs can write the papers in the students’ voices and in ways that are not detectable. This ability will be embedded in everyday products they use from laptops (even sans internet connections), to phones, and other devices they will carry. The job market for writers is starting to crater.

*Lots of blue links are not the future of search. Students are still being taught how to use Google in school, but that’s not the way they will search for information in the future. Even Google acknowledges that. They will use semantic searching in tools such as Perplexity.

Ex-Walmart exec Lore: The search engine is going to be archaic. It’s going to be the cassette tape in 20 years. Younger generations are going to laugh at the idea of using a search engine, because search engines aren’t that intelligent. 

Perplexity.ai is already starting to steal traffic from Google.

*AIs can already translate languages between people in nearly real-time when people are wearing earpieces, talking on the phone, and working in virtual meeting rooms. The job market for translators is starting to crater.

*The problem of AI hallucinations will decline radically, enabling bots to take on customer service roles with more confidence over the long term. The job market for customer service agents is starting to crater.

*The gap between students who use AI properly and those who don’t will widen. This isn’t just about bans, don’t-ban, but how students who use AI properly are running up the academic scorecard.

This is a great example of how a high school student is using AI.

It’s not hard to see who is going to have jobs in the AI World. It’s not going to be students who are “scared anti-AI straight.”

It’s students who attend the school run by a headmaster who DM’d me today with an invite to a program and said, “I lead a school committed to embracing AI in all we do.”

But you said I can keep up…This is crazy!

The way you can keep up is simple: Stop worrying about every potential AI development and start thinking about how to prepare your students to thrive in both school and life when machine intelligence is competitive with human intelligence.

That’s not simple, but that’s the North Star. The details of every AI development are irrelevant. And knowing the details isn’t relevant if you aren’t preparing your students for this world.

What’s the good news? Is that all you have?

The good news is that while most schools don’t know a lot about AI, they do know a lot about how to teach students how to think critically, communicate, collaborate, develop positive senses of self, and strengthen themselves as leaders. These are the most important skills in the AI-World.

From there, they can begin to learn about AI, changes to consider in their schools, and how to start teaching students with and about AI.

If you are an educational leader, it is time to get started. If you need help, ask.

Our paper will help you do that.