Thank you for sharing your Agentic AI explorations, Eric. I'm a high school English teacher that's been grappling with how to adjust to AI for a couple of years now. Like you say in this piece, the more I learn and explore, the more I'm left with even more questions.
Here's a few I'm thinking about now after reading your post:
What ethical and privacy concerns should we have about giving Agentic AI access to our drives, email, and calendar? What's the incentive for these companies to create these tools? Even money generated from premium level subscriptions doesn't come close to covering costs, so these tools at this point are effectively loss leaders, right? At some point if not already, our personal data or emerging dependency on these tools will be monetized. What will that look like in education or our personal lives?
AI slop and humanity
In a world where anyone can effortlessly create a convincing "Jeff Rogers," how do we know what or who is real and what isn't? And as you note, how can we identify real humans or real learning when fakes are so easy to generate?
Thanks again for your post and for sharing your thoughts and work on AI and education.
This is a nice tour of these new integrative capabilities, along with some thoughtful questions. I'm not sure about the term "agentic" here, but like the term "artificial intelligence," I guess we're stuck with a marketing term.
My concern, which comes through in some of the essay's questions, is the assumption that it is perfectly fine for these companies to roll out tools without understanding the scope or intensity of the problems they may bring. Asking "What's the worst that could happen?" leads to a shrug of the shoulders instead of a series of checks before the product is released.
The folks at The Verge have been arguing that these tools will be the death of what remains of the Web. This may be fine, except it is unclear whether AI model interfaces to the internet can function without important pieces of the Web continuing to exist. And then there is security. Some of the more technically minded writers I follow seem to have significant concerns about the risks associated with sensitive user information when it comes to these agentic capabilities.
If "AI as primary internet interface" succeeds, the long-term picture feels pretty grim, at least for those of us who fondly remember what it was like to go online in the 90s. Is there an emoji yet for "eyeroll at nostalgic old people saying how great things were when they were young"?
I wonder if writing classes might evolve into AI and non- AI versions. The AI version concentrates on curation and communication, finding patterns across objects/artifacts. The non-AI version features thinking & deep reading, obviously more in line with pre-AI times. Different approaches, different outcomes.
What you're seeing is is pattern completion, not planning, and it has none of the robustness of a planning solution.
However, there is some fun stuff here, and the Agentic workflows are better than RPA in some ways, but I can't get past the prompt injection opportunities. This will be a gold mine for anyone who listens to the little devil on their shoulder.
The entire internet is now your attack surface. Every website your agent visits, every email it reads, every form it parses are all good opportunities to siphon both personal information and cash from you.
The era of Agentic AI marks a fundamental shift from AI as a "chatbot" you talk to, to AI as an "agent" that does work for you. While traditional AI is reactive, waiting for a prompt to generate text or images, agentic AI is proactive—it can reason, plan, and execute multi-step tasks autonomously to achieve a specific goal.
I found the article to be informative and valuable. While testing the prompts and describing the outputs was useful, the questions you posed were the best part.
The big question is what is the role of teachers, students and schools in the face of Agentic AI and its rapid evolution? While some may find the technology to be exciting and seductive, it is more than disruptive in my view. It is destructive. The biggest question is what is the role of humanity in world of increasingly powerful AI “agents”? 8-10 people (tech trillionaires) own all the algorithms and are replacing humanity. AI is more productive, efficient, and obedient (so far).
When human thinking, meaning making, problem solving are best done by superior “beings”, what is the purpose and joy of being human? And let’s not think for a moment that the billionaire techbros are going to subsidize our food, shelter, health when robotic AI is doing all the work. We serve no purpose for corporate AI. 💥
Thank you for sharing your Agentic AI explorations, Eric. I'm a high school English teacher that's been grappling with how to adjust to AI for a couple of years now. Like you say in this piece, the more I learn and explore, the more I'm left with even more questions.
Here's a few I'm thinking about now after reading your post:
What ethical and privacy concerns should we have about giving Agentic AI access to our drives, email, and calendar? What's the incentive for these companies to create these tools? Even money generated from premium level subscriptions doesn't come close to covering costs, so these tools at this point are effectively loss leaders, right? At some point if not already, our personal data or emerging dependency on these tools will be monetized. What will that look like in education or our personal lives?
AI slop and humanity
In a world where anyone can effortlessly create a convincing "Jeff Rogers," how do we know what or who is real and what isn't? And as you note, how can we identify real humans or real learning when fakes are so easy to generate?
Thanks again for your post and for sharing your thoughts and work on AI and education.
Facinating and thought-provoking as always, Eric. Thank you for sharing your thoughts
This is a nice tour of these new integrative capabilities, along with some thoughtful questions. I'm not sure about the term "agentic" here, but like the term "artificial intelligence," I guess we're stuck with a marketing term.
My concern, which comes through in some of the essay's questions, is the assumption that it is perfectly fine for these companies to roll out tools without understanding the scope or intensity of the problems they may bring. Asking "What's the worst that could happen?" leads to a shrug of the shoulders instead of a series of checks before the product is released.
The folks at The Verge have been arguing that these tools will be the death of what remains of the Web. This may be fine, except it is unclear whether AI model interfaces to the internet can function without important pieces of the Web continuing to exist. And then there is security. Some of the more technically minded writers I follow seem to have significant concerns about the risks associated with sensitive user information when it comes to these agentic capabilities.
If "AI as primary internet interface" succeeds, the long-term picture feels pretty grim, at least for those of us who fondly remember what it was like to go online in the 90s. Is there an emoji yet for "eyeroll at nostalgic old people saying how great things were when they were young"?
I wonder if writing classes might evolve into AI and non- AI versions. The AI version concentrates on curation and communication, finding patterns across objects/artifacts. The non-AI version features thinking & deep reading, obviously more in line with pre-AI times. Different approaches, different outcomes.
What you're seeing is is pattern completion, not planning, and it has none of the robustness of a planning solution.
However, there is some fun stuff here, and the Agentic workflows are better than RPA in some ways, but I can't get past the prompt injection opportunities. This will be a gold mine for anyone who listens to the little devil on their shoulder.
The entire internet is now your attack surface. Every website your agent visits, every email it reads, every form it parses are all good opportunities to siphon both personal information and cash from you.
The era of Agentic AI marks a fundamental shift from AI as a "chatbot" you talk to, to AI as an "agent" that does work for you. While traditional AI is reactive, waiting for a prompt to generate text or images, agentic AI is proactive—it can reason, plan, and execute multi-step tasks autonomously to achieve a specific goal.
Agentic AI is a game changer for the job industry and all other industries, that's why its best to learn it early.
Discover the top Agentic AI developments: https://promptengineer-1.weebly.com/agentic.html
Also, access a library of over 2000 prompts, refer to, Agentic AI Prompt Vault: https://promptengineer-1.weebly.com/agentic-ai-prompt-vault.html
I found the article to be informative and valuable. While testing the prompts and describing the outputs was useful, the questions you posed were the best part.
The big question is what is the role of teachers, students and schools in the face of Agentic AI and its rapid evolution? While some may find the technology to be exciting and seductive, it is more than disruptive in my view. It is destructive. The biggest question is what is the role of humanity in world of increasingly powerful AI “agents”? 8-10 people (tech trillionaires) own all the algorithms and are replacing humanity. AI is more productive, efficient, and obedient (so far).
When human thinking, meaning making, problem solving are best done by superior “beings”, what is the purpose and joy of being human? And let’s not think for a moment that the billionaire techbros are going to subsidize our food, shelter, health when robotic AI is doing all the work. We serve no purpose for corporate AI. 💥
This article really made me think, connecting so well with your earlier observations on AI's evolvin' impact; agentic capabilities are simply amazing!
+1 on prompt injection. Terrifying prospect. Interesting read here if anyone encounters this and wants to read more: https://brave.com/blog/unseeable-prompt-injections/