A Summary of the Exeloop? Dinner Discussion | Philadelphia, October 14, 2026

Scientific discovery increasingly depends on capabilities beyond any single laboratory, discipline, or institution. Advances in biological measurement, computational science, and artificial intelligence let researchers investigate relationships that were difficult to explore even a few years ago. But the institutions, incentives, and practices supporting research have not necessarily changed at the same pace.

The research ecosystem is also under strain. Federal funding is uncertain, early-career investigators face a competitive environment, valuable data remain hard to combine, and discoveries face long paths to clinical use. Questions about reproducibility and public trust in science add to these pressures.

To explore these changes, the Karen Toffler Charitable Trust (KTCT) hosted a private Exeloop? dinner in Philadelphia on October 14, 2026, bringing together leaders from academia, research, philanthropy, biotechnology, clinical and translational science, law, commercialization, and investment. The discussion examined how science is changing, what keeps promising discoveries from advancing, and what future collaborations may be needed.

Science Is Crossing Boundaries Faster Than Institutions Are

Many of today's most difficult scientific questions no longer fit within a single discipline or institution. Understanding complex biological systems requires combining expertise, technologies, and data traditionally developed separately.

Neurodegenerative disease provides a clear example. Researchers confront mixed pathologies, interactions among biological systems, population differences, and disease processes that may begin decades before symptoms appear. Investigating these relationships can require neuroscience, immunology, genetics, computational science, clinical research, and longitudinal patient data.

AI is beginning to help researchers generate hypotheses, design experiments, and predict biological behavior. Advances in multi-omics, spatial biology, and human-based models are revealing relationships previously difficult to observe. Yet participants questioned whether researchers can evaluate and use these technologies as quickly as they develop.

These capabilities can accelerate hypothesis creation and reveal relationships researchers might otherwise miss. But computational predictions are not experimental evidence. New insights still require biological validation, reproducibility, and the judgment of scientists who understand the limitations of the models and data being used.

Developments beyond the dinner offer examples of new approaches. Biohub's Virtual Biology Initiative combines biological research, large-scale data generation, AI, and computing to develop predictive models of biology, drawing on capabilities across organizations.

These developments are changing what scientists can investigate, but not necessarily the institutional arrangements governing how they work together.

When the Relationship Exists but the System Gets in the Way

Participants questioned whether money or access to collaborators is always what prevents promising research from advancing.

Researchers may know who has the data, models, or expertise they need but face months of negotiation over access, ownership, publication rights, or intellectual property. Academic institutions and companies may share scientific objectives while operating under different incentives and expectations.

Many protections are legitimate. Universities must safeguard intellectual property, companies must protect assets, patient data require privacy safeguards, and researchers deserve recognition. But individually reasonable protections can collectively create obstacles.

When institutions cannot agree on ownership of discoveries not yet made, negotiations over hypothetical future value can delay the research needed to determine whether that value exists. In trying to protect their individual interests, institutions may prevent everyone involved from learning whether there is anything worth protecting.

Trust matters as much as formal agreements. Collaborators need confidence that their contributions will be recognized, data will be used responsibly, and legitimate interests will be respected. Agreements establish rules, but relationships and behavior determine whether people can work together when outcomes remain uncertain.

Participants also questioned the assumption that government research funding is necessarily free money. Programs such as Small Business Innovation Research (SBIR) can deliver valuable non-dilutive support, but awards carry conditions. Spending restrictions, reporting and compliance requirements, and other obligations can add costs or constrain how a company operates. The question is not simply how much funding is available, but what accepting it requires and whether those conditions serve the research or the organization pursuing it.

These concerns raise broader questions. Which protections are essential? Where could standard agreements replace repeated negotiations? What could be treated as precompetitive? And when do the conditions attached to funding or collaboration begin to undermine their intended benefits?

The challenge is not to eliminate institutional interests. It is to recognize when protecting them reduces the collective value they could create.

New Models of Collaboration Are Emerging

Texas's Dementia Prevention and Research Institute of Texas (DPRIT), established with a $3 billion research fund, is intended to support dementia research, prevention, treatment, commercialization, talent, and collaboration.

Although implementation has faced legal delays, participants described how Texas universities have been preparing for future funding opportunities by bringing researchers together and supporting interdisciplinary teams.

Other initiatives take different approaches. Aligning Science Across Parkinson's organizes multidisciplinary teams around shared objectives and expectations for data sharing. ARPA-H and Wellcome Leap bring capabilities together around defined problems rather than relying exclusively on traditional investigator-driven funding.

These models show that funding and institutional design can shape not only what research is pursued, but also how researchers work together.

Building the Connective Infrastructure

Biomedical research produces enormous quantities of genomic, proteomic, imaging, biomarker, and clinical data. Their value increasingly depends on whether they are representative, interoperable, accessible, and usable. NIH's Bridge2AI program tackles these challenges through AI-ready datasets and the standards and tools needed to make them useful.

But technical infrastructure is only part of the answer.

Scientific progress also depends on people who can interpret findings, understand technical limitations, recognize unexpected relationships, and determine what to investigate next.

The ecosystem needs cognitive infrastructure: the accumulated knowledge, judgment, relationships, and experience people use to make sense of complex problems.

It resides in the researcher who understands why an experiment failed, the clinician who recognizes when a patient does not fit the expected pattern, and the investigator who sees a connection across fields. Much of this knowledge cannot be captured in a database or transferred through a publication.

Technology can generate information at an accelerating rate, but the human capacity to evaluate and act on that information cannot expand as quickly.

When researchers leave, laboratories shrink, or programs disappear, the loss is not simply headcount. Accumulated experience and relationships disappear with them.

A dataset or computational platform can sometimes be restored when resources return. The knowledge and judgment developed through years of scientific work cannot be rebuilt so easily.

Early-career researchers develop scientific judgment through experiments, failed hypotheses, interpretation, and mentorship. As AI begins taking on tasks traditionally performed by junior researchers, institutions will need to consider whether greater efficiency is reducing opportunities to develop that judgment. Researchers must still learn to question AI-generated results, recognize the limitations of the tools producing them, and test findings against experimental evidence. Greater efficiency today should not come at the expense of developing the scientific judgment needed tomorrow.

New Infrastructure Requires New Behaviors

A more connected research system also requires changes in how institutions reward and support scientific work.

Universities may need to simplify agreements. Researchers may need to share data, methods, or negative results earlier. Companies may need to distinguish genuine competitive advantages from shareable resources. Funders may need to recognize contributions beyond individual grants and publications..

Participants questioned whether existing recognition and reward systems discourage the very collaborations scientific progress increasingly requires.

If a researcher contributes essential expertise to a project but is not the principal investigator or lead author, how is that contribution recognized? If building a shared resource advances an entire field but produces fewer publications, how is that work valued?

Leadership is equally important. Scientific collaborations increasingly require people who can recognize connections across disciplines, understand different institutional interests, and bring expertise together around shared problems.

Leading such collaborations requires different skills. Bringing talented researchers together does not automatically create an effective team. Leaders must establish shared objectives, navigate competing interests, build trust, recognize contributions, and make decisions amid uncertainty.

If collaboration is becoming essential to scientific progress, institutions and funders must treat the ability to lead it as something worth developing and supporting.

Trust must also be reinforced through agreements, incentives, leadership responsibilities, and commitments to shared resources. Sometimes creating greater collective value requires giving up some control.

Where Discovery Meets Translation

Participants discussed the difficulties of moving scientific discoveries toward clinical use. Biomedical innovation is often described as a pipeline, but the capabilities and evidence required at each stage differ considerably.

A promising finding may rely on a model that inadequately reflects human disease. Validation may involve populations unlike those who will eventually receive a treatment. Data collected for scientific purposes may not satisfy the requirements of regulators, clinicians, companies, or payers.

Reproducibility is particularly important. A finding may be compelling enough to publish but difficult to reproduce under different conditions. Experimental methods, biological models, and incomplete supporting data can affect whether another organization can confidently build upon it.

This is also a question of trust. Researchers, companies, and clinicians need evidence they can examine, test, and rely upon.

Public confidence in science depends on similar principles, but rebuilding trust requires more than transparency and communicating scientific findings. Scientists and institutions must also understand the experiences and concerns that shape people's perceptions of science and engage with them respectfully. Questions and skepticism should not automatically be treated as a failure to understand science. People need to understand how scientific knowledge develops and why findings sometimes change, but they also need to know their concerns are being heard and taken seriously. That includes being candid about what has not worked, where evidence remains incomplete, and why changing scientific conclusions can reflect progress rather than failure. Trust cannot be rebuilt simply by explaining science better. It requires listening as well as informing.

Participants considered whether translation could improve if different expertise entered the research process earlier. Industry can help identify evidence needed for development. Clinicians can highlight differences between experimental models and patients. Patient organizations can inform research priorities and study design. Regulatory expertise can identify validation requirements before critical decisions.

This does not mean every discovery should begin with a commercial endpoint. Curiosity-driven research remains essential because its applications cannot always be anticipated. Researchers pursuing fundamental scientific questions must retain the freedom to explore without being required to identify a commercial use. The challenge is to introduce translational perspectives when useful without prematurely narrowing scientific research.

Participants also emphasized allowing people to do what they do best. Exceptional discovery scientists may not be the right people to lead commercialization, just as experienced entrepreneurs may not possess the expertise required for basic research.

The emerging model is less a one-way pipeline than a connected system in which clinical, scientific, regulatory, and commercial knowledge inform one another.

Instead of asking only how to move discoveries across the valley of death, what knowledge needs to move upstream so fewer promising discoveries reach the valley unprepared?

Philanthropy Can Fund the Space Between Institutions

Philanthropy can fund the space between institutions, supporting work that falls outside traditional funding mechanisms or requires organizations to collaborate before they have sufficient evidence or shared experience to pursue larger grants.

It cannot replace government funding for basic biomedical research or match industry's capital once a technology demonstrates commercial potential. But philanthropic organizations can support unconventional ideas, provide continuity for early-career researchers, and help establish relationships and capabilities that larger funding systems may not support.

A modest grant might help investigators test a shared idea, make datasets compatible, access specialized expertise, or involve clinicians and industry partners earlier. Such investments may produce preliminary evidence, attract additional collaborators, or position a team for larger funding.

Participants also considered philanthropy's role in developing scientific talent and leadership. Mentorship, cross-disciplinary training, and opportunities to work in different institutional environments can build capabilities whose value goes beyond individual projects.

As larger initiatives such as DPRIT create new funding opportunities, smaller philanthropies may have a distinctly valuable role in preparing researchers and teams to pursue them.

In that environment, leverage may matter as much as scale.

Philanthropy can also serve as a neutral convener. Universities, companies, investors, and government agencies have different interests and obligations. Philanthropy can create opportunities for those organizations to explore shared problems without requiring them to abandon their individual interests.

An equally important measure of impact may be what an investment makes possible next.

What Must Change Now

The discussion pointed to a future in which scientific progress will increasingly depend on relationships that don't naturally exist today. Researchers will need capabilities beyond their disciplines. Universities and companies will need to work across different incentives. Funders will need to support contributions that extend beyond individual projects. Institutions will need to develop and retain people who can connect knowledge across these boundaries.

These relationships must also extend beyond the scientific community. Rebuilding public trust will require scientists and institutions not only to communicate what they know, but to listen to and respect the people whose lives their work is intended to improve.

The collaborations science will need five or ten years from now will not all emerge on their own. Building them requires decisions today about what institutions reward, protect, share, and sustain.

The question is not simply what collaborations the future will require. It is what we need to do differently now to make them possible.