What Is Science as a Service? Bringing Scientific Expertise to Industry and Society
A concept for bringing measurement, physics-based models, and AI to practical challenges in industry and society.
Read the articleAI × SCIENCE × PHYSICAL WORLD
AI for
Science.
Rethinking how we do science with AI.
AI Scientists specialized for each research domain connect observation, reasoning, execution, and verification.
Articles↓NUCLEAR + PARTICLE EXPERIMENTS
NUCLEARAI SCIENTIST
Coding for nuclear and particle experiments
From detector signals and simulation code to execution, verification, and feedback.
QUANTUM IMAGING
QUANTUMAI SCIENTIST
Quantum imaging
Reconstruct what cannot be seen directly. Let measurement, inference, and experiment speak together.
LUNAR LIFE + SYSTEMS
LUNARAI SCIENTIST
Living on the Moon
A scientific agent for living systems beyond Earth: environment, resources, energy, and operation.
FUSION REACTOR CONTROL
FUSIONAI SCIENTIST
Fusion reactor control
Physics, prediction, control, and feedback in one closed loop.
GENERAL PHYSICS LAW DISCOVERY
DISCOVERYAI SCIENTIST
Discovering physical laws
Data becomes hypothesis. Hypothesis becomes model. Models survive only by being tested.
Rethinking how we do science with AI.Not one model. A constellation of scientific agents.
Articles↓What Is AI for Science? How AI Is Changing Scientific Research
Explore how AI connects observation, prediction, and experiment, and why scientific verification matters.
Read the articleVision
From AI as a tool
to AI as part of
the scientific method.
We see AI as part of the scientific method, connecting observation, hypotheses, computation, experiment, and verification within a scientific discovery loop.
SCIENTIFIC INTELLIGENCE STACKconcept
05FeedbackUpdate from results
04VerificationCheck against reality
03ExecutionRun computations and experiments
02ReasoningForm hypotheses and plans
01ObservationGather data and observations
RESEARCH PROGRAMDesigning intelligence
Designing intelligence
for science.
We see AI as an intellectual foundation connecting discovery, computation, experiment, and verification.
Find patterns in data and turn them into hypotheses.
Model the physical world computationally and make its predictions testable.
Connect experiments and measurements to AI reasoning loops.
AI Scientists iterate through planning, execution, verification, and improvement.
Research areas
One question.
Multiple ways to explore it.
We study AI as an intellectual foundation connecting discovery, computation, experiment, and verification. Our research explores how to extend the process of scientific discovery while preserving the expertise of each domain.
01
Scientific Discovery
We study how AI can support scientific discovery through an iterative exchange between data, hypotheses, and models.
02
Scientific Computing
Combining simulation and machine learning to make complex scientific computations faster and easier to work with.
03
Experimental Intelligence
Connecting measurement and experimental data with AI to turn signals from the physical world into scientific understanding.
04
Scientific Agents
We explore AI agents for scientific research that repeatedly plan, execute, verify, and improve their work.
CONCEPT SPACEResearch ideas,
Research ideas,
made tangible.
Closed-loop
Intelligence.
Intelligence that traces results back to their causes and updates the next hypothesis and action.
Many sciences.
One loop.
Connecting scientific domains through a shared structure of observation, reasoning, execution, and verification.
Grounded
in reality.
People ask questions, AI explores, and the physical world provides evidence.
Concept Space
Research ideas,
before they become papers.
Our central research question is how to build intelligence for doing science. The concepts below visualize this approach.
01AI that learns
CLOSED-LOOP INTELLIGENCE
AI that learns
from consequences.
Go beyond generation: trace execution results back to their causes, then update the next hypothesis and action.
AISCIENTIST
OBSERVE
REASON
ACT
VERIFY
LEARN
02Different domains.
MANY SCIENCES, ONE LOOP
Different domains.
Shared intelligence.
Nuclear science, quantum imaging, lunar living, fusion, and law discovery: different domains can share an intelligence architecture that connects observation to verification.
03Science stays
HUMAN × AI × PHYSICAL WORLD
Science stays
grounded in reality.
People formulate questions, AI accelerates exploration, and the physical world provides answers. Science advances through this exchange.
AI SCIENTISTSSwipe the
Swipe the
frontiers.
A family of specialized AI Scientist concepts grounded in the scientific constraints and tools of each domain.
01 / 05
Coding for nuclear and particle experiments
Nuclear & Particle Experiment Coding Agent
An AI Scientist concept spanning simulation, analysis code, and data processing in experimental research, with support for execution, verification, and improvement.
02 / 05
Quantum imaging
Quantum Imaging Agent
An AI Scientist concept connecting reconstruction, noise analysis, and experimental design in quantum measurement and imaging to extract information from observations.
03 / 05
Living on the Moon
Lunar Living Agent
An AI Scientist concept bringing together science and engineering to consider the environment, resources, energy, and operations involved in living on the Moon.
04 / 05
Fusion reactor control
Fusion Reactor Control Agent
An AI Scientist concept using both physical models and data to connect plasma understanding, prediction, and control support in a closed loop.
05 / 05
Discovering physical laws
General Physics Law Discovery Agent
An AI Scientist concept that develops hypotheses from data, generates mathematical models, and explores candidate physical laws through simulation and attempts to falsify them.
AI Scientists
Different sciences.
A shared intelligence architecture.
We explore how to build specialized AI Scientists that understand the tools, constraints, and verification methods of each scientific domain, connecting those domains through a shared architecture of observation, reasoning, execution, and verification.
From coding and measurement
to control and discovery.
AI SCIENTIST / LIVE CONCEPT SPACE01 — 05
Coding for nuclear and particle experiments
Nuclear & Particle Experiment Coding Agent
An AI Scientist concept spanning simulation, analysis code, and data processing in experimental research, with support for execution, verification, and improvement.
Quantum imaging
Quantum Imaging Agent
An AI Scientist concept connecting reconstruction, noise analysis, and experimental design in quantum measurement and imaging to extract information from observations.
Living on the Moon
Lunar Living Agent
An AI Scientist concept bringing together science and engineering to consider the environment, resources, energy, and operations involved in living on the Moon.
Fusion reactor control
Fusion Reactor Control Agent
An AI Scientist concept using both physical models and data to connect plasma understanding, prediction, and control support in a closed loop.
Discovering physical laws
General Physics Law Discovery Agent
An AI Scientist concept that develops hypotheses from data, generates mathematical models, and explores candidate physical laws through simulation and attempts to falsify them.
Science as a Service / ScaaS
Bringing AI × Physics
to society and manufacturing.
Science as a Service (ScaaS) is an initiative that combines knowledge of physics with AI, measurement, and data processing to support practical problem solving. We use data to understand challenges, develop hypotheses, and connect them to implementation and verification.
Putting AI × Physics into practice
We connect physical models and simulations with AI to design and implement methods for analysis and prediction. We verify them against both data and phenomena, exploring approaches suited to each challenge.
Addressing societal challenges with data
We organize observational, experimental, and operational data, using visualization and analysis to understand the structure of a problem. Testing hypotheses helps inform decisions and actions toward a solution.
Supporting manufacturing
through measurement and data processing
We connect the evaluation of measurement methods with signal processing, analysis, and interpretation. Understanding phenomena and variability supports problem solving and improvement in manufacturing.
EDUCATIONLearning Intelligence
Learning Support Agent
Connecting learning history
to what comes next.
We explore educational agents that support each learner's current needs by using learning history as context: past coursework, understanding, engagement with assignments, difficulties, and progress.
The aim is to use learning history over time to adapt explanations, assignment difficulty, next topics, and feedback. We seek a useful layer of support for learners and teachers while preserving teachers' educational judgment.
EDU
ConceptSupport informed by learning history
Learning History-Aware Support Agent
01Learning historyCoursework, understanding, activities
02Current stateEstimate current understanding
03Adaptive supportAdapt explanations, tasks, and cognitive load
04FeedbackGuide the next learning step
PASTNOWNEXT
CourseworkUnderstandingAssignmentsDifficultiesCurrent stateAdaptive supportNext steps
LEARNING CONTEXTUse learning history as context for deciding what support comes next.
Context
Consider the learner's history alongside the current question.
Adaptation
Adapt explanations, assignments, and pace to understanding and cognitive load.
Agency
Support learners while preserving their agency and teachers' judgment.
Privacy
Define why and how learning history is used, and handle only the information needed.
Scientific discovery loop
Science is not a prompt.
It is a loop.
High-quality scientific agents need feedback that helps trace results back to their causes, allowing hypotheses and actions to be updated continuously.
01ObserveObserve phenomena
02HypothesizeDevelop hypotheses
03SimulateTest computationally
04ExperimentTest in the physical world
05LearnUpdate from results
Human ↔ AI ↔ Physical World
Principles
Designed for science,
not for demos.
These principles guide the development of work that can be verified scientifically.
01
Physics firstKeep scientific questions and physical consistency at the center of the work.
02
Closed loopTreat generation, reasoning, experiment, and feedback as one connected research loop.
03
Human agencyPeople formulate questions, verify findings, and retain responsibility for the research and its decisions.
Publications & Research Works
A record of research.
Research works
Public research works registered in ORCID, listed by publication year, newest first.
103 public research works
NEWEST FIRSTApplication of β-NMR to spectroscopy and imaging
Nuclear Science and Technology
Performance of prototype dual gain multilayer thick gas electron multiplier with high-intensity heavy-ion beam injections in low-pressure hydrogen gas
Progress of Theoretical and Experimental Physics
Pole position of Λ(1405) measured in d(K−,n)πΣ reactions
Physics Letters, Section B: Nuclear, Elementary Particle and High-Energy Physics
“K−pp” a K‾-meson nuclear bound state, observed in 3He(K−,Λp)n reactions
Physics Letters, Section B: Nuclear, Elementary Particle and High-Energy Physics
Structure near the K- + p + p threshold in the in-flight 3He(K-, Λp)n reaction
Progress of Theoretical and Experimental Physics
Addendum to "measurement of the Li 8 (α,n)11 B reaction and astrophysical implications"
Physical Review C - Nuclear Physics
Hypernuclear production cross section in the reaction of 6Li + 12C at 2A GeV
Physics Letters, Section B: Nuclear, Elementary Particle and High-Energy Physics
Search for the deeply bound K- pp state from the semi-inclusive forward-neutron spectrum in the in-flight K- reaction on helium-3
Progress of Theoretical and Experimental Physics
Search for deeply bound Kaonic nuclear states via 3He(K-, n) reaction at J-PARC
Chiral Symmetry in Hadrons and Nuclei
Search for the K- pp bound state via the 3He(K-, n) reaction at 1 GeV/c
Journal of Physics: Conference Series
High-power beam dump system for the BigRIPS fragment separator at RIKEN RI Beam Factory
Nuclear Instruments & Methods in Physics Research Section B-Beam Interactions with Materials and Atoms
Search for a bound kaon and pion state
Progress of Theoretical and Experimental Physics
BigRIPS separator and ZeroDegree spectrometer at RIKEN RI Beam Factory
Progress of Theoretical and Experimental Physics
Direct measurement of the 18Ne(α,p)21Na reaction
AIP Conference Proceedings
The K1.8BR spectrometer system at J-PARC
Progress of Theoretical and Experimental Physics
Gas-gain study of standard CERN GEM and 400-mu m-thick Thick GEM in low-pressure He/CO2 mixed gas
Nuclear Instruments & Methods in Physics Research Section a-Accelerators Spectrometers Detectors and Associated Equipment
THE SEARCH FOR DEEPLY BOUND KAONIC NUCLEAR STATES AT J-PARC
International Journal of Modern Physics A
A search for deeply bound kaonic nuclear states at J-PARC
AIP Conference Proceedings
Development of GEM chamber using E10 experiment
Proceedings of the Sendai International Symposium: Strangeness in Nuclear and Hadronic Systems, SENDAI 2008
Development of the GEM-MSTPC for measurements of low-energy nuclear reactions
Nuclear Instruments & Methods in Physics Research Section a-Accelerators Spectrometers Detectors and Associated Equipment
Hypernuclear spectroscopy with Heavy Ion Beams: The HypHI project at GSI and FAIR
Proceedings of the Sendai International Symposium: Strangeness in Nuclear and Hadronic Systems, SENDAI 2008
HYPERNUCLEAR SPECTROSCOPY WITH HEAVY ION BEAMS: THE HypHI PROJECT AT GSI AND FAIR
International Journal of Modern Physics E-Nuclear Physics
Identification of 45 New Neutron-Rich Isotopes Produced by In-Flight Fission of a U-238 Beam at 345 MeV/nucleon
Journal of the Physical Society of Japan
Production of neutron-rich lambda hypernuclei at J-PARC
Proceedings of the Sendai International Symposium: Strangeness in Nuclear and Hadronic Systems, SENDAI 2008
PRODUCTION OF NEUTRON-RICH LAMBDA HYPERNUCLEI AT J-PARC
International Journal of Modern Physics E-Nuclear Physics
A systematic study of astrophysical reaction rates through 8Li
AIP Conference Proceedings
Scintillating fiber detectors for the HypHI project at GSI
Nuclear Instruments & Methods in Physics Research Section a-Accelerators Spectrometers Detectors and Associated Equipment
THE HYPHI PROJECT: HYPERNUCLEAR SPECTROSCOPY WITH STABLE HEAVY ION BEAMS AND RARE ISOTOPE BEAMS AT GSI AND FAIR
International Journal of Modern Physics E-Nuclear Physics
Direct measurement of 8 Li + d reactions of astrophysical interest
AIP Conference Proceedings
EXPADES: A new detection system for charged particles in experiments with RIBs
Nuclear Instruments & Methods in Physics Research Section B-Beam Interactions with Materials and Atoms
In-beam tests of scintillating fibre detectors at MAMI and at GSI
Nuclear Instruments & Methods in Physics Research Section a-Accelerators Spectrometers Detectors and Associated Equipment
Magnetic field-map measurement of a superconducting triplet quadrupole and a dipole magnet for BigRIPS separator at RIKIEN
Ieee Transactions on Applied Superconductivity
Magnets for the beam transport line in the RIKEN RI beam factory
IEEE Transactions on Applied Superconductivity
Magnets for the beam transport line in the RIKEN RI beam factory
Ieee Transactions on Applied Superconductivity
Kaon-Nucleus interaction studied through the in-flight (K-, N) reaction
Progress of Theoretical Physics
Gated multiple-sampling and tracking proportional chamber new detector system for nuclear astrophysical study with radioactive nuclear beams
Nuclear Instruments & Methods in Physics Research Section a-Accelerators Spectrometers Detectors and Associated Equipment
Production of a low-energy radioactive nuclear beam with high purity using JAERI-RMS
Nuclear Instruments & Methods in Physics Research Section a-Accelerators Spectrometers Detectors and Associated Equipment
Low-energy radioisotope beam separator CRIB
Nuclear Instruments & Methods in Physics Research Section a-Accelerators Spectrometers Detectors and Associated Equipment
Prototype of superferric quadrupole magnets for the BigRIPS separator at RIKEN
Ieee Transactions on Applied Superconductivity
Subbarrier fusion in the systems 11,10Be+209Bi: The experimental data
AIP Conference Proceedings
Design features of the in-flight RI beam separator BigRIPS at RIKEN
International Meeting on Nuclear Applications of Accelerator Technology: Accelerator Application in a Nuclear Renaissance
Spin-parity assignments in 15C by a new method: β-delayed spectroscopy for a spin-polarized nucleus
Physical Review C - Nuclear Physics
Measurement of the 8Li(α,n)11B reaction and astrophysical implications
Physical Review C - Nuclear Physics
Multiple-sampling and tracking proportional chamber for nuclear reactions with low-energy radioactive isotope beams
Nuclear Instruments & Methods in Physics Research Section a-Accelerators Spectrometers Detectors and Associated Equipment
Performance of fragment separator with homogeneously thick energy degrader: A highly flexible system for use with degraders in a wide thickness range
Nuclear Instruments & Methods in Physics Research Section a-Accelerators Spectrometers Detectors and Associated Equipment
Radioactive core ions of microclusters, ''snowballs'' in superfluid helium
Hyperfine Interactions
Radioactive core ions of microclusters, "snowballs" in superfluid helium
Hyperfine Interactions
No matching works found. Try different keywords or filters.
Records are grouped using public ORCID information and matching DOIs. The count includes work types other than journal articles. Records without a DOI are not merged solely on the basis of similar titles.
Direction of travel
From specialized agents
to a scientific intelligence ecosystem.
Starting with AI Scientists specialized for individual domains, we aim to identify shared feedback structures and verification principles and develop an intelligence foundation that can extend across scientific research.
01
Domain AgentsNuclear science, quantum imaging, lunar living, fusion, law discovery
02
Shared Feedback ArchitectureReasoning, execution, verification, feedback that traces causes
03
Closed-loop ScienceA research ecosystem in which AI, people, and the physical world continually learn from one another