Setting the Stage: Jack Hidary and Sandbox AQ
Decades of Research, Limited Progress
Alzheimer's, Parkinson's, dementia, and cancer have seen decades of research with few effective treatments, highlighting the urgent need for new approaches.
Instead of the world of large language models, we've now entered the world, Peter, of large quantitative models, LQMs.
Jack Hidary's Journey
Why AI and Quantum? The Nexus of Information Compression
Grand Challenges in Medicine and Energy
Despite decades of research, Alzheimer’s, Parkinson’s, pancreatic cancer, and glioblastoma remain untreatable. Energy transition progress is halting. These massive challenges demand new tools.
Steve Jobs died of pancreatic cancer, billions in the bank account, nothing to do.
The AI–Quantum Nexus
Both AI and quantum technology model the world by compressing enormous datasets into manageable, actionable insights, enabling predictions and outputs to tackle grand challenges.
Large Language Models: Strengths and Limitations
Evolution of Language Models
LLMs as compression engines
A large language model compresses a huge text corpus down to its essence — capturing patterns like “carness” — in the weights of the network, allowing it to generate coherent outputs.
There is no equation for the English language.
Beyond Words: The Quantitative World of Numbers
The Quantitative Nature of Reality
Most real-world challenges, from drug design to battery chemistry, are governed by numbers and physical laws, not by language. Language models alone cannot solve these problems without precise mathematical understanding.
the majority of our world, Peter, is not words, but numbers.
Quantum Equations Underpinning Physical Laws
From Newton to Quantum: Equations as Ultimate Compressors
Equations as Ultimate Compressors
Newton's laws compress macro-scale dynamics into simple equations, while quantum equations compress subatomic behaviors, enabling precise modeling from rocket trajectories to drug design.
Everything is Quantum.
Key Quantum Milestones
Modeling Drug Targets with Quantum Precision
Beyond Classical Protein Modeling
While AI tools like AlphaFold predict static protein folding, SandboxAQ leverages quantum equations to model dynamic electron interactions—critical for designing drugs like cancer immunotherapies that target protein receptors.
Evolution of Protein Modeling
And that's the level now, finally, that we at SandboxAQ have been able to model things at. And that is a big breakthrough.
The Breakthrough: Solving Quantum Equations on GPUs
GPUs Solve Quantum Equations Today
The same GPU infrastructure that enabled large language models can accurately solve classical quantum equations—Schrödinger, Bohr, Planck, Heisenberg—without requiring a scaled quantum computer. This allows immediate modeling of atoms, electrons, and ions, unlocking insights across health, materials, and the environment.
How Large Quantitative Models (LQMs) Generate Pristine Training Data
- Start with a target molecule (e.g., a drug for glioblastoma).
- Create a digital twin of that molecule.
- Generate millions to billions of chemical permutations (adding/removing methyl, amine, nitrogen groups, etc.).
- Solve quantum equations on GPUs for each permutation to produce a clean, physics-based dataset.
- Train AI models on this generated data to predict which variants will effectively hit the target.
LLMs vs. LQMs: Where the Data Comes From
So, first there we take the quantum equations, we run those, and that becomes the data set. So, we're generating our data set and that's what we use to train the AI.
LQMs in Action: From Simulations to Real-World Impact
Large Quantitative Models (LQMs)
Instead of relying on messy real-world data, LQMs start with the pristine quantum equations of the universe, generate accurate synthetic data, and train AI to tackle everything from drug discovery to new battery chemistries.
How LQMs Power Real-World Solutions
- Start with the quantum equations as the bedrock
- Generate diverse synthetic data by creating variations on themes (e.g., molecules, ions)
- Train large quantitative models on this pristine data
- Apply models to design new drugs, biomarkers, batteries, and materials
Revolutionizing Drug Development and Clinical Trials
Quantum-Enabled AI Tackles Molecular Complexity
By focusing only on the valence electrons at the business end of a molecule, SandboxAQ makes molecular modeling tractable on today's GPUs, unlocking a new era of quantitative AI for drug discovery.
Transforming Drug Development Metrics
90% failure today in clinic. 90 out of 100 drugs that go into clinical phase one trial, phase two, phase three, never see the light of day.
The Road Ahead for Quantum Computing
Quantum Computing Milestones
Current Qubit Technologies
Neutral atoms, a rapidly scaling dark horse, use lasers to manipulate atoms near absolute zero, offering room-temperature operation and compact form factors. Photonics leverages silicon photonics and semiconductor fabs for mass production, with key players like PsiQuantum and Photonic.
How a Qubit is Used
- Initiate a state on the qubit (e.g., using lasers to set a neutral atom into 0, 1, or a superposition).
- Perform a set of operations (gates) on the qubit states.
- Read out the final state at the end of the computation.
they can also be in superpositions, in combinations of zero and one. And that gives us an infinite palette to draw from.
Information: The Unifying Principle
Quantum Sensing is Already Here
Unlike quantum computers, quantum sensors require no error correction or millions of qubits; they are operational today in planes for navigation when GPS is denied and in hospitals for magnetocardiography.
Information: The Unifying Principle
Both AI and quantum physics rely on summarizing large amounts of data or complexity into compact representations: neural networks learn compressed representations, while physics condenses dynamics into equations.
This fundamental insight that information is the building block of our universe.
Parting Thoughts: Questions over Answers
If we are in a simulation, then the beings who created the simulation, kudos to them. They've done a pretty good job.
Questions Spark Breakthroughs
The Power of Questions
Contributions to society can take the form of not just answers, but the questions we pose to ourselves and the next generation.
Let's focus on the questions and not just the answers.