At the end of an AI training I was teaching, I told the room:
“Don’t be scared of AI.”
There were three of us teaching that day, and we covered a lot. Tools, prompts, ways to use AI for work and business, and a lot of information in a short amount of time.
By the end, I knew it was a lot to take in.
What I meant was simple.
You do not have to remember everything you heard today.
If you get home and forget how to do something, ask AI.
If a tool feels confusing, ask AI to walk you through it one step at a time.
If the answer is too much, ask it to make it simpler.
You can use AI to help you learn how to use AI.
But that is not exactly what I said.
I said, “Don’t be scared of AI.”
And those two things are not the same.
Why that matters to me
I often do better in writing than I do when I am speaking.
I’m autistic, and live conversation can move faster than I can fully process what I want to say. With writing, I can read my words, reread them, notice that they do not quite say what I meant, and change them.
Speaking does not always give me that time.
Sometimes the sentence that comes out is close to the thought in my head, but it is missing a piece.
That is what happened that day.
“Don’t be scared of AI” can sound like “There is nothing about AI to be scared of.”
That is not what I meant.
There are real reasons to pay attention
In September 2026, AI researcher Jacob Coxon resigned from Anthropic and warned that leading AI companies were moving too fast toward more powerful systems without enough attention to safety. He had previously worked at OpenAI.
Around the same time, Evan Hubinger, Anthropic’s Alignment Science Lead, said his personal estimate was that there is a greater than 10% chance AI could kill all humans within the next decade.
That is his estimate, not an official Anthropic prediction.
There is also serious disagreement about what recent AI incidents mean.
Scientific American reported that some security researchers see recent cases of AI agents getting into systems they should not have accessed as major security and oversight problems, but not necessarily evidence that an extinction-level outcome is coming. Their focus is on stronger controls, monitoring, and security practices.
I do not know what the long-term risk will turn out to be.
But we do not need to know that answer to see that AI already raises real questions.
NIST, the National Institute of Standards and Technology, already tracks risks such as false or misleading output, privacy problems, harmful bias, information integrity, and people relying too heavily on AI systems.
There are also questions about jobs, power, and how much control a small number of companies may have over technology that is becoming part of everyday life.
Learning still matters
Even with all of that, I still think people should learn how to use AI.
Not because AI is harmless.
Because it is here.
The people in that training were trying to understand something new. After listening to three speakers cover a lot of information, I did not want them leaving feeling like they had to remember every prompt, every tool, or every step.
They could keep learning after the training ended.
They could ask questions.
They could experiment.
They could ask AI to explain something in simpler words.
And they could question the answers it gave them.
I still think that matters.
Learning how to use AI does not mean trusting everything about AI.
You can use it and question it.
You can find it useful and still care about privacy, safety, jobs, power, and where the technology may be heading.
You can learn something new and change your mind as you learn more.
If I were teaching that training again, I would still want people to leave feeling like AI was something they could learn.
I would just finish the thought.
Don’t be scared of AI.
Pay attention to it.