This week, the Center for Humane Technology published a powerful piece about the dangers of consumer-facing AI systems. Read it here. It describes the tragic case of a patient who committed suicide after interacting with an AI that was designed—like much of today’s technology—for maximum engagement, not maximum wellbeing.
This case should stop us in our tracks. We are repeating the same mistakes made with the internet and social media: building systems optimized for addiction, usage, and stickiness, instead of human flourishing. When profit models are tied to time spent rather than health gained, the inevitable outcome is harm.
At TruNeura, we believe the right use of AI in healthcare is exactly the opposite: AI should not be designed to replace the doctor or compete for patient attention—it should be built to empower the clinician, support the patient’s daily healing, and track progress in a way that creates real-world health outcomes in an emerging clinical area where no medication has ever worked.
The Wrong Path: AI Built for Addiction
Optimized for engagement, not outcomes. Just like social media algorithms, many consumer AI systems are designed to keep people coming back, regardless of whether that time is healthy or destructive.
Illusion of intimacy. AI chatbots can feel personal but lack context, training, and responsibility. This false intimacy can be dangerous when patients are in crisis.
No clinical accountability. When there’s no physician oversight, there’s no feedback loop to ensure what the AI recommends is safe, evidence-based, or effective. Recommending suicide is about the most clear case of this impending disaster.
This is why we see AI-powered health tools “in the wild” will likely end up causing harm rather than healing.
The Right Path: AI Built for Healing
At TruNeura, we’re designing AI to work with doctors, not around them.
Here’s what that looks like in practice:
Daily guidance for patients. The AI checks in with patients every day—asking if they are doing the small, evidence-based behaviors that prevent and reverse cognitive decline. It reminds them of their personalized goals, keeping them engaged in the healing process, not in addictive loops.
Synthesis for clinicians. Instead of drowning in data, the physician gets a clear, synthesized picture of what’s driving the pathology—from labs to lifestyle—so they can focus on healing.
Tracking real outcomes. AI follows the patient’s progress over time, showing both patient and doctor what’s working and where to adjust.
Learning from success. The AI compares each patient’s journey to others who have successfully reversed cognitive decline, guiding them along the most effective, cost-efficient path.
This is not AI that replaces the physician, but AI that elevates them—helping them spend less time on data wrangling and more time in the healing relationship.
Why This Matters Now
The tragic case shared by the Center for Humane Technology is a warning: if we let AI follow the path of social media, we will deepen the crises of mental illness, addiction, and disconnection.
But if we commit to building AI systems that are accountable, physician-centered, and outcome-driven, we can flip the script. Instead of fueling despair, AI can become a trusted ally in the fight against chronic illness—and especially in cognitive decline, where daily consistency and clinician insight make all the difference.
At TruNeura, we’re taking a risk by doing it differently. We’re betting that the future of healthcare AI isn’t more clicks—it’s more healing.
Curious if your AI has an internal “world model” like a chess bot would, or if it explicitly doesn’t like the LLMs? I’m borrowing this question from Gary Marcus’s work - https://garymarcus.substack.com/p/generative-ais-crippling-and-widespread