In my first post, I shared a story about my late husband worrying that GPS devices would rob our boys of the essential skill of reading a map. I remember wondering—even back then—how essential that skill really was. Would they need it for their chosen careers? …their everyday lives? …or only in the event of an apocalypse?
That memory returned to me as I’ve been exploring today’s AI learning tools.
Because once again I find myself asking:
Are these new AI tools truly “new,” or are we just putting lipstick on a pig? :>
And just as importantly:
What do these tools give us—and what might they take away?
Is AI Innovation… or Evolution?
The truth, I suspect, is both.
Just today, Google launched its newest model, Gemini 3, described as its ‘most intelligent model’ yet—integrated into Search, the Gemini app and enterprise platforms from day one.
Google claims benchmark-leading scores (e.g., tops LMArena) and improved reasoning, multimodal ability and ‘agentic’ workflows. But even here the fine print matters: independent commentary notes that while Gemini 3 shows strong improvement, factual accuracy remains about ~72 % on some benchmarks—meaning that even ‘state-of-the-art’ still has significant error-rates. This example becomes a prompt for both hope (the leap) and realism (still error-prone).
Are we gaining genuinely new capabilities, or just a higher-powered version of what we already had (and possibly trading off things like simplicity, trust, accessibility)?
Some AI-driven platforms are simply expanded versions of older tools we already understand—Google Jamboard becoming Miro, Otter.ai doing 10x what transcription tools used to do, Canva adding “Magic Write” to a familiar interface.
Other tools feel like true game-changers: automated grading systems, generative video creators like Synthesia, adaptive microlearning through 7Taps, and personalized pathways from platforms like Nolej or PrepAI.
But even with these breakthroughs, we must stay honest about limitations.
AI still reflects the data it consumes. And biased data leads to biased outputs—something both Harvard’s implicit bias research and countless AI audits have made painfully clear.
Humans are biased. Machines are biased. The difference is scale. A biased human affects a classroom; a biased algorithm affects an entire institution.
So the question becomes:
How do we adopt new AI tools responsibly, while acknowledging both their power and their pitfalls?
A Working Lens: Where Did These Tools Come From, What Are They Now, and What Do They Add—or Take Away?
I’ve begun work on a chart that begins mapping the evolution from pre-AI tools to their more AI-enabled successors. This isn’t meant to be comprehensive—it’s an invitation to reflect. It’s visible here. To enlarge the chart, please use the +/zoom tool. Please feel free to add constructive comments.
You’ll notice that some AI tools add meaningful capabilities. Some simply automate what we already do. And some raise real ethical or accessibility concerns, particularly around privacy, data usage, bandwidth requirements, and cost. (Download Full Chart .pdf)
Ethical Implications: The Part We Cannot Ignore
The promise of AI in learning is enormous:
more time for feedback
personalized pathways
better accessibility (Mercury 13 for low vision, AI captioning, etc.)
But AI also raises serious concerns:
Bias amplification
Data privacy risks
Cost and bandwidth inequities
Even a major player like Google admits non-perfect accuracy in Gemini 3, raising questions about blanket trust in “new AI” tools, especially in educational contexts and for learners with different needs. Responsible instructional design means:
using encryption and anonymizing learner data
deleting data when no longer needed
conducting regular audits
staying aware of who gets left behind when tools require expensive hardware or high-speed internet
We cannot design ethically if we don’t design critically.
So… Lipstick on a Pig, or Something Truly New?
Maybe the healthiest answer is curious skepticism.
AI tools are evolving quickly—but they are evolving from something.
And if we understand their lineage, we can better evaluate both their promise and their cost.
I would love to hear from others:
What older tools do you see reflected in today’s AI platforms?
What do you feel we’re gaining—and what do you worry we might be losing?