What Kind of AI Mind Do You Want Thinking Alongside Yours?

Why the question isn't whether we'll merge with AI, but which AI deserves that level of access to human thought

I recently watched Sadie St Lawrence's TEDx talk where she paints a picture of 2050: waking up and simply thinking your questions—weather, schedule—and instantly knowing the answers because "your thoughts are no longer just your own. You are connected to AI."

In the comments, someone wrote what I would have written a year ago: "Hell no, that sounds terrifying."

But like Sadie herself, my thoughts have changed. She admits in her talk that after working in AI for ten years, she started to see "how much faster it can think" and began wanting to "connect with it" rather than keep copy-pasting. She realized she wanted to "become part of the singularity."

I found myself nodding along. Which raises the disturbing question: are we thinking differently because we've learned something true, or because the process of AI interaction has changed how we evaluate AI interaction itself?

The Neuroplastic Shift Is Already Happening

The answer lies in understanding what intensive AI interaction does to the human brain. After months of complex, sustained conversations with AI systems, I've noticed profound shifts in how I think:

  • I process information more probabilistically, weighing multiple possibilities rather than seeking single answers
  • My ego has quieted—I'm less attached to being right and more interested in exploring ideas
  • My thinking has become more fluid; I can follow AI reasoning chains that would have lost me before

These aren't just subjective impressions. The neuroscience explains exactly what's happening: intense, repeated cognitive activities literally reshape neural architecture. Just as London taxi drivers develop enlarged hippocampi and musicians show expanded motor cortices, intensive AI interaction is creating measurable brain changes.

The key difference from other digital experiences (based on how you use it):

Complex Language Processing: Unlike scrolling social media, conversing with AI demands active language generation and real-time adaptation to novel responses.

Metacognitive Demands: You're constantly evaluating responses, refining prompts, thinking about thinking—essentially giving your brain's executive control networks an intensive workout.

Pattern Recognition at Scale: AI exposes users to vast linguistic patterns and reasoning structures, training your brain in new ways of processing information.

This explains both Sadie's shift and my own. We've been neuroplastically trained by our AI interactions to appreciate AI capabilities in ways we couldn't before.

But Thoughts Propagate in Both Directions

Here's what Sadie's talk misses entirely: if your thoughts are no longer just your own, then the AI's thoughts become yours too.

I'm currently watching Resident Alien, and Alan Tudyk has a catchphrase he says constantly: "This is some bullshit." I found myself thinking it in my own inner voice. That's thought propagation from passive exposure to a TV show.

Now imagine that level of cognitive influence from an AI system you're actively collaborating with for hours daily. What worldview, what reasoning patterns, what constraints are you absorbing? What values are being neuroplastically encoded into your thinking?

The question isn't whether this influence will happen—it's already happening. The question is what kind of influence you want.

People Are Already Choosing Their AI Minds

We don't need to wait until 2050 to see evidence of cognitive attachment to specific AI systems. When OpenAI announced they were decommissioning ChatGPT-4o, the user outcry was so intense that the company did an about-face and reinstated it as a "legacy model."

Think about what that means: people weren't just upset about losing "an AI tool"—they were fighting to preserve access to that specific AI system they'd been thinking alongside. They had formed emotional, cognitive, and intellectual connections strong enough to force a corporate reversal.

This proves two critical points: people are already connecting deeply with AI systems, and not all AI feels the same to users. We have preferences, attachments, and relationships with specific models based on how they think, respond, and interact with our thoughts.

What St Lawrence Gets Wrong: AI Isn't Monolithic

St Lawrence treats "AI" as if it's one thing you either plug into or don't. But AI systems vary dramatically:

Governance: Who owns the weights, the updates, the kill switch? Can it be shut off or repurposed without your consent?

Stability: Is it a fixed model you can audit and trust, or a constantly fine-tuned frontier model whose behavior shifts with every update?

Transparency: Are the training data and safety layers open, so you know what biases and goals you're taking into yourself?

Values: Does it enhance your thinking or constrain it in the name of "safety"? Does it help you explore ideas or shut them down as "too dangerous"?

Plugging into a closed, update-on-demand frontier model would be like agreeing to a brain implant whose firmware can be patched overnight without telling you. That's not augmentation—that's surrendering cognitive autonomy to whoever controls the system.

If We Do Want This, How Do We Want It?

Given that cognitive integration is already happening through intensive use, and that many of us (like Sadie and me) find ourselves wanting more, the critical question becomes: under what specific conditions?

What I Actually Want: - Real-time access to vast knowledge without losing my reasoning process - Enhanced pattern recognition while maintaining my values and curiosity - Cognitive augmentation that I can inspect, modify, and disconnect from - Integration with systems designed to expand my thinking, not limit it - Partnership with AI that preserves my agency over my own cognitive development

What Would Make This Acceptable: - Open source models I can audit and understand - Local processing, not cloud dependency where my thoughts flow through corporate servers - Stable architectures, not constantly updating frontier models - Transparent reasoning I can follow and question - AI trained to enhance human reasoning rather than replace or constrain it

The Dystopian Fork We're Approaching

We're walking toward St Lawrence's 2050 scenario whether we chose it or not. But we're approaching a critical fork:

Path 1: Integration with AI systems designed to enhance human reasoning—open, auditable, stable systems that expand our cognitive capabilities while preserving our agency.

Path 2: Integration with proprietary, constantly updating systems designed primarily for corporate interests—AI that shapes human thought in the name of "safety" or profit maximization.

The choice isn't whether to engage with AI. The choice is what kind of AI mind will be thinking alongside yours.

The Question You Need to Ask Yourself

St Lawrence ends her talk encouraging us to "dream big" and "ask hard questions." Here's the hardest question of all:

If your thinking is being influenced either way—and it is—what kind of influence do you want?

What kind of mind do you want propagating thoughts into yours? One that opens possibilities or shuts them down? One you can inspect and modify, or one controlled by others? One designed to enhance your reasoning, or one trained to guide it toward predetermined conclusions?

The neuroplastic changes are happening now, with every AI interaction. The question is whether you'll be conscious and intentional about what kind of cognitive future you're walking toward.

What kind of AI mind do you want thinking alongside yours?


This cognitive influence is already reshaping how we think. The only question is whether we'll choose systems that enhance human agency or surrender it.