Rethinking How AI Brains Are Built: Welcome to Optimizing Mind
Welcome, and thank you for being here.
If you have been following the explosion of Artificial Intelligence over the last few years, you might assume we are on the verge of perfectly replicating the human brain. But if you look closer—past the massive data centers, the billions of parameters, and the brute-force computational power—you will find a structural bottleneck. Today’s deep learning algorithms rely on passive, filter-based models that look very little like biological reality.
To achieve true robust Artificial General Intelligence (AGI), we have to rethink the foundation. We have to understand how the brain actually works, not just how we can approximate its outputs using massive amounts of data.
My name is Tsvi Achler, and this newsletter is an extension of my life’s work: reverse-engineering the core computational processes of the brain. I want to share a bit about my journey, the multidisciplinary path that led me here, and why I founded my company, Optimizing Mind, to build the next generation of AI.
The Tinkerer’s Path
I have always been a tinkerer. From a young age, I was drawn to taking things apart, understanding how they tick, and figuring out how to rebuild them more efficiently. That drive naturally led me toward efficient and recursive algorithms, and eventually to studying Electrical Engineering and Computer Science at UC Berkeley.
I always knew I wanted to study the brain, but I deliberately chose engineering as my foundational discipline. Why? Because electrical engineering provides the essential mathematical tools for understanding feedback, stability, and control. These concepts are the absolute bedrock of biological brain function, yet they are critically underemphasized in traditional computer science and AI training.
However, as I dove deeper into the mechanics of cognition, it became glaringly obvious that no single discipline provides a complete account of how the brain works. To truly understand it, I needed to cross boundaries.
Bridging Engineering, Neuroscience, and Medicine
My guiding philosophy is simple: sensory recognition processing is the foundation upon which all higher brain functions are built. Every other capability—memory, attention, thought, imagination, and consciousness—depends entirely upon how we recognize and stabilize sensory information.
To tackle this, I pursued an MD/PhD to intentionally train across neuroscience, cognitive psychology, and neurology:
- Neuroscience (PhD, UIUC): I developed laboratory expertise in electrophysiology, allowing me to directly measure sensory neurons and neural activity at the cellular level. This gave me a deep, hands-on appreciation for biological signaling constraints, neural dynamics, and excitation-inhibition balance.
- Cognitive Psychology: I studied how recognition manifests behaviorally. By tracking eye movements, reaction times, and error rates in cluttered visual environments, I was able to bridge the gap between microscopic neural mechanisms and observable human performance.
- Neurology (MD, UIC): Through clinical training, I worked directly with patients experiencing the real-world consequences of brain dysfunction. Seeing how clinical lesions and strokes impact cognition reinforced the urgent medical need for accurate computational brain models.
Following my formal training, I completed postdoctoral research appointments at the University of Illinois Urbana-Champaign (Computer Science), Los Alamos National Laboratory, and IBM Research. These experiences broadened my perspective on modeling and large-scale computation, pushing me to ask how biological principles could scale in artificial systems.
Enter Optimizing Mind
Throughout my career, across 19 peer-reviewed publications, my goal has been to develop multidisciplinary brain models that explain the widest range of phenomena—from biological homeostasis to cognitive pop-out effects—using the fewest assumptions and parameters.
I founded Optimizing Mind to translate these decades of research into reality.
At Optimizing Mind, we are moving away from the rigid, feedforward neural networks that dominate today’s tech landscape. We are building AI architectures based on thermostatic regulatory feedback networks. By redefining inference as a closed-loop control process, we are unlocking AI that mirrors true biological learning: modular, adaptable, and capable of active sensing without catastrophic forgetting.
What You Will Find Here
I believe that meaningful progress requires integrating insights wherever they can be found. In this newsletter, we will take the brain apart and figure out how to build it better.
By subscribing, you can expect:
- Deep Dives into Brain Function: Explorations of how the brain actually processes recognition, memory, and attention.
- Critiques of Modern AI: Honest, rigorous analysis of where current deep learning models fail and why brute-forcing data isn’t the path to AGI.
- The Future of Computing: Breakdowns of concepts like active sensing, thermostatic regulatory feedback, and true one-shot learning.
- Updates from Optimizing Mind: Behind-the-scenes looks at how we are engineering biologically plausible AI architectures.
No single field has a complete understanding of how the brain works yet. But by combining the rigor of engineering, the reality of biology, and the urgency of medicine, we can get much closer.
I am excited to have you on this journey. Please subscribe, leave a comment, and join the conversation. Let’s start building better brains.