No Adult Left Behind: What AI Can Do—and Where Humans Still Matter
Author: George Pillari
AI is learning. But are we?
That is the question sitting behind George Pillari’s No Adult Left Behind, a practical, industry-oriented book that tries to move AI away from the “either you are a tech person or you are doomed” narrative.
The book is divided into three broad parts—Productivity, Careers, and Industries—and its central argument is fairly straightforward:
AI can amplify what humans do, but humans still have to do the thinking.
And that brings us back to Stanislav Petrov.
In 1983, the Soviet early-warning system incorrectly indicated that the United States had launched nuclear missiles. Petrov, the officer on duty, judged the warning to be a false alarm rather than blindly passing it up the chain. Later investigation confirmed that the system had malfunctioned.
Different technology, different era, same uncomfortable question:
What happens when the machine is wrong and the human stops questioning it?
AI Productivity and Career Skills: From Prompting to Presentations
The first two sections of No Adult Left Behind are highly practical.
The Productivity section looks at choosing the right AI tool for the right task, prompting, search, chatbots, agents and everyday problem-solving. The Careers section moves into resumes, interview preparation, professional development and the skills humans still need in an AI-assisted workplace.
Some of this will sound familiar if you have already read several AI books:
AI is a powerful tool, not a replacement.
Garbage in = garbage out.
Good prompts need task, context, format and constraints.
AI can be an excellent assistant but a troublesome decision-maker.
Iterate until you get the result you actually need.
Upskill instead of waiting for AI to pass you by.
So, nothing new?
Not quite.
The book has some genuinely useful additions. AI as an editor is one. Tailoring a master resume for different roles is another. The STAR-based approach to structured prompting is worth exploring, and the discussion of interview preparation goes beyond simply asking AI to conduct another predictable mock interview.
The PPT section was particularly useful to me. Guy Kawasaki’s 10-20-30 rule, Canva, Gamma and Tome, one central idea per slide, and the distinction between a point and a topic are practical rather than theoretical.
In the world of PPTs, I am a problem-facer too. So yes, I took notes.
The “Thank You” Problem: AI Efficiency Needs Responsible Examples
But AI books also need to think about the examples they teach us.
One example asks AI:
“Thank you, but can you go deeper on that last point?”
By the way, shouldn’t we drop the “Thank you” if the stated objective is to reduce unnecessary tokens and conserve computational resources?
Yes, I understand that it is only an example. But that tiny sentence raises a bigger question:
Should an AI book model responsible AI behaviour as carefully as it recommends it?
This is also where the book’s repeated advice about permissions, preferences, iterative prompting and human supervision becomes important.
AI can be patient. It can be non-judgemental. It can keep generating possibilities long after a human colleague has started wondering whether lunch is more productive.
But patience is not judgement.
Amazon’s Rufus, for example, is designed to answer shopping questions, compare products and make personalised recommendations using Amazon’s catalogue, reviews, community Q&As and information from across the web. Amazon itself acknowledges that generative AI will not always get things exactly right.
That is useful assistance.
It should not automatically become the final decision.
Titanic, Iron Dome and the Ethics of Automation
The Industries section is where the book becomes considerably more interesting.
Its examples include Titanic, Space Shuttle Challenger, and, US Airways Flight 1549, alongside questions involving autonomous cars, medicine and military systems.
The Titanic discussion particularly stayed with me. The idea of aggregating scattered information into a fuller picture is powerful. If decision-makers had possessed better integrated information, would every decision have changed?
We cannot know. But the example demonstrates something important:
Information is useful only when it reaches the person who can interpret and act on it.
The military examples make the human-machine relationship even more complicated.
Israel’s Iron Dome, for instance, incorporates substantial automation, but its engagement process includes operator intervention. Israeli missile-defence officials have described operators as able to override system recommendations.
That is precisely where the Human-in-the-loop versus Human-on-the-loop distinction stops being academic.
And Anthropic provides another fascinating contemporary example. In 2026, the company said it would not knowingly provide its products for fully autonomous weapons that remove humans from target selection and engagement, arguing that current frontier AI is not sufficiently reliable for that role and that proper safeguards are necessary. At the same time, Anthropic explicitly continues to support other national-security applications.
So the real question is not simply “Can AI do it?”
It is “Should AI be allowed to decide it?”
The Missing Conversation: When Augmentation Becomes Dependence
This is my biggest reservation about No Adult Left Behind.
The book makes a strong case for augmentation. But I wanted a deeper discussion of what happens when augmentation gradually becomes dependence.
We already know that humans can become over-reliant on automated systems. We know that AI can misread information, hallucinate, misidentify and confidently produce an answer that sounds perfectly reasonable.
And that takes us right back to Petrov.
The lesson of Petrov’s story is not that humans are magically better than machines. The warning system detected something. The human assessed it, questioned it and refused to treat the machine's output as unquestionable truth.
That principle matters just as much when AI helps write a resume, select a product, prepare a presentation or assist with a high-stakes industrial decision.
So, what should we do?
Use AI. Experiment with it. Learn it. Upskill. Let it challenge your thinking.
But do not outsource your judgement along with your workload.
Let AI become your assistant, editor, researcher, simulator and occasionally your very patient problem-solving partner.
Just keep the human in the loop.
After all, AI is learning.
Are you?
Comments