The Missing Member of Every AI Team: The Public

What if the next breakthrough in medicine fails, not because the science is wrong, but because nobody trusts the people bringing it to market?

Biology and computation are rapidly becoming one research system. Machine learning can search chemical space, interpret biological signals, automate experiments, and coordinate the development process through shared digital infrastructure. The old image of a laboratory team as a biologist working beside a chemist is giving way to a three person model: biologist, chemist, and computer scientist.

That is a profound change in how medicines may be discovered. It is also an incomplete one.

The missing member is the public.

Not as a passive audience, a focus group, or a population whose data can be harvested. The public is a participant in the system whose trust determines whether a technical possibility becomes a social reality. The central challenge for the next generation of biotechnology is therefore not merely integrating life sciences with computer science. It is integrating technical power with public legitimacy.

The future of technology will not be decided only by what can be built. It will be decided by what people are willing to believe, permit, use, and defend.

The convergence is real, but convergence creates a legitimacy gap

Consider what happens when computation enters drug discovery. A software system may identify a promising molecule from patterns no human researcher could easily see. A cloud platform may let laboratories share results, tools, and workflows at unprecedented speed. Open software infrastructure may make parts of research reproducible and accessible to more teams.

These developments can compress years of work into months. They can lower the cost of experimentation and make sophisticated biological analysis available to smaller companies. They can also change who controls knowledge, whose judgment counts, and where accountability sits when a decision is made by a chain of models, databases, automated experiments, and investors.

The technical system becomes more powerful precisely as it becomes harder for an outsider to understand.

That is where the public response to generative AI offers an important warning. People do not necessarily reject invention, automation, or scientific progress. Many people reject the social arrangement surrounding innovation. They see executives becoming wealthy while workers absorb the disruption. They see products introduced without consent, cultural practices treated as obstacles, and companies demanding exceptional legal and political treatment. The resentment aimed at AI is often resentment toward the institutions that deliver it.

This distinction matters enormously for biotech. A person might welcome a treatment discovered with machine learning while distrusting the company that owns the model, the hospital that gathers the data, or the investor who benefits from the resulting patent. A patient may support faster drug development while objecting to opaque use of medical records. A researcher may value automation while fearing that the tools will turn professional judgment into a cost center.

The technology and its context cannot be separated in practice, even if they can be separated philosophically.

A useful analogy is food. A person can love bread and distrust a factory that conceals its ingredients, underpays its workers, and contaminates a river. Saying that the customer should distinguish the product from the producer misses the point. Production methods affect the meaning and safety of the product itself.

The same is true of computational biology. Governance is not packaging added after the science. It is part of the science's social operating environment.

The hidden fourth discipline is legitimacy

The three person team of biologist, chemist, and computer scientist solves an important technical problem. Each discipline sees a different layer of reality. Biology explains living systems. Chemistry manipulates matter. Computer science helps find structure in complexity and coordinates the work.

But a medicine is not successful merely because it is biologically effective. It must also be trusted by patients, prescribed by clinicians, reimbursed by institutions, approved by regulators, and accepted by communities. It must enter a network of human decisions.

This suggests a more complete model:

1. Discovery: Can we identify a promising intervention?
2. Validation: Does it work reliably in biological and clinical settings?
3. Translation: Can it be manufactured, distributed, and used safely?
4. Legitimation: Do the people affected understand its basis, trust its stewards, and regard its deployment as fair?

Most technology companies treat the fourth stage as public relations. That is a strategic mistake. Legitimacy is not a mood surrounding the product. It is a productive capacity that determines adoption, data access, recruitment for clinical trials, regulatory cooperation, and long term survival.

A company can have excellent models and still fail at the legitimacy stage. It can discover a candidate drug but lose access to the patient communities whose participation is required to test it. It can build an impressive health platform but trigger resistance because people cannot tell who owns their data. It can develop a highly accurate diagnostic system that clinicians refuse to use because no one can explain its errors when a patient is harmed.

The technical team therefore needs a fourth role: not necessarily one person with a formal title, but a permanent function responsible for the relationship between innovation and the public.

Call this role the social systems engineer. This person asks questions that are neither purely technical nor merely ethical:

• Who bears the risks if this system fails?
• Who benefits if it succeeds?
• What must users understand before they can consent meaningfully?
• Which communities are supplying the data, labor, or biological material?
• Can an affected person challenge a decision, correct an error, or withdraw participation?
• What promises can the company make that will still be credible after the next funding round?

These questions do not slow innovation by definition. They prevent the kind of failure that appears late, after capital has been spent and trust has been lost.

Why medicine may become the test case for technology's reputation

The public image crisis surrounding AI is especially instructive because generative AI is often encountered as an intrusion. A new system appears in a workplace, school, search engine, or creative field, and people are told to adapt. The benefits are described in universal language, while the costs arrive as individual obligations: learn a new tool, accept surveillance, compete with automated output, or absorb a reduction in status.

Biotechnology can reproduce the same pattern in a more intimate domain. Health data is not just another digital resource. It contains family histories, genetic probabilities, stigmatizing diagnoses, reproductive information, and evidence of vulnerability. If people sense that their bodies are being converted into assets for distant institutions, even genuine medical benefits may be interpreted as extraction.

The danger is not simply that people will misunderstand complex science. It is that they will understand the underlying bargain all too well.

A company may say that data is being used to accelerate cures. The public may hear: your most private information will be collected, analyzed, and monetized, while you receive no durable control over the system. A company may say that automation democratizes research. Scientists may hear: the value of your expertise will be captured by a platform whose owners set the terms. A company may say that an algorithm improves diagnosis. Patients may ask: who is accountable when it is wrong?

These are not irrational objections. They are questions about power.

The more technologically sophisticated a system becomes, the more important it is to make its power legible. People do not need to understand every line of code, just as airline passengers do not need to understand the physics of lift. They do need to know who is responsible, what safeguards exist, what happens when something goes wrong, and whether they have any meaningful recourse.

This is the difference between explanation and accountability. An explanation tells people how a system works. Accountability tells them who must answer for its consequences. Advanced companies often offer the first while avoiding the second.

A new mental model: innovation as a trust pipeline

The usual innovation pipeline runs from research to product to market. A more realistic model adds trust at every stage. Think of it as a series of gates, each of which can preserve or destroy legitimacy.

Gate one: origin

Where did the data, samples, ideas, and labor come from? If a company cannot answer this clearly, its scientific advantage may conceal a moral liability. Provenance should be treated as a design requirement, not an archival detail.

Gate two: participation

Who gets to influence the questions being asked? A system built entirely around what investors find commercially attractive may miss the problems patients consider urgent. Participation does not mean allowing every stakeholder to veto research. It means recognizing that affected communities possess knowledge about needs, risks, and practical constraints.

Gate three: interpretation

Can the important conclusions be challenged? In medicine, a model's output should not become an unquestionable oracle. Clinicians and patients need ways to inspect evidence, compare alternatives, and understand uncertainty. A confident prediction is not the same thing as a justified decision.

Gate four: distribution

Who receives the benefits, and at what price? A treatment discovered efficiently but priced beyond the reach of the people whose data enabled it may be scientifically successful and socially defective. Distribution is part of the product's design, not a problem for someone else after launch.

Gate five: repair

What happens after failure? Trust does not require perfection. It requires visible mechanisms for correction. Companies that disclose errors, compensate harms, revise models, and accept external scrutiny can retain credibility. Companies that hide behind technical complexity turn every mistake into evidence of bad faith.

This pipeline yields a practical principle: the public should not be asked to trust a technology at the end of its development if trust was absent from the beginning.

That principle has implications for organizations trying to unite technology and life sciences. They should build public accountability into funding decisions, product architecture, clinical strategy, and executive incentives. A governance review should not be a ceremonial document prepared before launch. It should have the power to change which questions are pursued, which data are used, and which markets are considered acceptable.

What builders can do now

The solution is not to make every technologist perform public relations, nor to substitute popular opinion for expertise. The solution is to create institutions in which technical excellence and social responsibility reinforce each other.

A research company working at the intersection of computation and biology can begin with several concrete practices:

• Publish a data constitution. State what kinds of data are collected, from whom, for which purposes, under what retention rules, and with what rights of withdrawal or correction.
• Create an independent challenge function. Give clinicians, patient advocates, security specialists, and domain experts a formal route to question assumptions before a system reaches the market.
• Measure legitimacy as an operating metric. Track consent quality, community participation, adverse event response time, explanation quality, and the distribution of benefits, not just model accuracy and research velocity.
• Design for recourse. Every consequential automated recommendation should have an identifiable human owner, an appeal process, and a documented method for correcting errors.
• Explain the economic bargain. Tell users who profits, how value is shared, and what obligations the company accepts in return for access to their data or participation.

These practices may appear expensive compared with simply launching a product. But the relevant comparison is not between governance and speed. It is between early investment in legitimacy and late stage resistance, litigation, regulation, failed trials, public backlash, or abandonment.

There is also a lesson for investors. Funding the entire journey of medicine development is more than a financial strategy. It creates responsibility for the full chain of consequences. Capital that supports discovery, software infrastructure, clinical translation, and commercialization cannot treat social failure as someone else's problem. If investors want to build the future, they must finance the institutions that make the future acceptable.

Key Takeaways

1. Treat public trust as infrastructure. It supports data access, clinical participation, adoption, and regulatory cooperation just as surely as cloud systems support computation.
2. Add legitimacy to the technical team. Biologists, chemists, and computer scientists need a durable social systems function that represents affected communities and accountability.
3. Design recourse before deployment. Users should know who is responsible, how errors are corrected, and how consequential decisions can be challenged.
4. Make the economic bargain visible. Explain who supplies the data and labor, who captures the gains, and what the institution owes in return.
5. Measure more than performance. A system is not ready when it is accurate alone. It is ready when its provenance, governance, distribution, and repair mechanisms are credible.

The deepest mistake in technology is to imagine that resistance comes from ignorance. Sometimes it does. More often, resistance is a judgment about character, incentives, and power. People may not hate the tools. They may hate being treated as raw material for someone else's progress.

The next great biotech companies will certainly need better models, better laboratories, and better software. But they will also need a different theory of innovation. The winning organization will not be the one that merely discovers the fastest. It will be the one that can connect discovery to a social contract people recognize as fair.

The future team is not made up of three technical specialists. It is made up of those three specialists plus the public, represented through real participation, transparent governance, and enforceable accountability.

That is not a concession to fear. It is the final piece of the technology itself.