Why We Need AI Standards — and Why Universities Must Lead the Charge

Artificial intelligence isn’t just another technology trend — it’s a transformative force reshaping the world. From healthcare to education, transportation to entertainment, AI systems are rapidly becoming woven into the fabric of everyday life. Yet despite its enormous potential, AI also poses profound ethical, societal, and technical challenges. If left unchecked, this technology could deepen inequities, amplify misinformation, threaten jobs, and even undermine safety and security on a global scale.

To navigate this transformation responsibly, we urgently need robust standards and guidelines for AI — the kind that have guided the development of the internet and the World Wide Web. And at the heart of this effort should be university AI research departments, whose independence, depth of inquiry, and commitment to societal good make them uniquely suited to steward this critical work.

The Problem: AI Without a Common Framework

Right now, AI is advancing faster than our ability to govern it. Companies build proprietary models with little transparency. Startups launch products without safety audits. Governments debate regulation while adversarial actors exploit loopholes. In this context, three things become painfully clear:

1. There’s no common baseline for what “good” AI looks like.
   Should systems be fair? Accountable? Transparent? Secure? How do we measure these properties in a meaningful way?
2. Different stakeholders have different incentives.
   Private firms focus on innovation and profit. Governments are fragmented by jurisdiction and politics. International coordination remains difficult.
3. Existing standards bodies (ISO, IEEE, etc.) are important but limited.
   They often move slowly and lack the interdisciplinary expertise needed to grapple with AI’s broader social implications.

This is precisely the gap that AI standards and guidelines — crafted collaboratively, openly, and globally — could fill.

What We Can Learn from the Web’s Standardization Journey

In the early days of the internet and the World Wide Web, innovation was rapid but chaotic. Different protocols competed. Compatibility was inconsistent. Without shared standards, the web could easily have fractured into siloed systems that couldn’t communicate with each other.

Enter the World Wide Web Consortium (W3C) — a collaborative international community that developed open standards (like HTML, CSS, and HTTP) to ensure the web remained interoperable, accessible, and scalable. Although not perfect, the W3C model offers a powerful blueprint for coordinating technical and ethical norms around a shared technology.

Here are two lessons we should take from the W3C experience:

1. Standards Encourage Innovation, Not Stifle It

A common misconception is that standards slow progress. In reality, they unlock innovation by:

• Eliminating redundancies (so engineers don’t reinvent the wheel)
• Enabling interoperability across platforms and tools
• Reducing barriers to entry for new developers and companies
• Providing a shared language for technical and ethical commitments

Just as HTML and CSS enabled the web to scale, AI standards can allow different models, tools, and platforms to work together safely and efficiently.

2. Openness Is Essential

The W3C’s open, consensus-based process allowed diverse voices — researchers, companies, governments, civil society — to contribute. This model reduced fragmentation and ensured that standards reflected a broader range of interests.

For AI, transparency and openness will be even more critical, given the technology’s societal implications.

What AI Standards Should Address

AI standards aren’t just about technical specifications. They must encompass ethical, societal, legal, and economic dimensions. A few key areas include:

1. Safety and Robustness

• Ensuring AI behaves predictably under varied conditions
• Preventing catastrophic failures in critical systems (e.g., autonomous vehicles, healthcare diagnostics)

2. Fairness and Non-Discrimination

• Defining metrics to detect and mitigate bias
• Establishing norms around equitable data collection and usage

3. Transparency and Explainability

• Setting expectations for when and how AI models should be interpretable
• Clarifying documentation standards for datasets and algorithms

4. Privacy and Security

• Guarding against misuse of personal data
• Protecting AI systems from adversarial attacks

5. Accountability and Governance

• Clarifying legal liability for AI outcomes
• Creating processes for audits, impact assessments, and redress

6. Environmental Impact

• Assessing and reducing the carbon footprint of training and deploying AI systems

These are not just technical challenges — they are social ones, too. Standards must balance innovation with responsibility, and that means bringing diverse perspectives to the table.

Why Universities Should Be at the Forefront

Many stakeholders could contribute to AI standards — corporations, governments, NGOs, international organizations. But universities have distinct strengths that position them as ideal leaders:

1. Independence and Public Trust

Universities are (ideally) less beholden to commercial incentives and short-term market pressures. This independence fosters credibility and public trust — essential ingredients for stewarding work that affects billions of lives.

While companies may be reluctant to share proprietary data or methods, universities can act as neutral conveners that bridge the interests of industry, government, and civil society.

2. Deep Interdisciplinary Expertise

AI isn’t just a technical field — it intersects with philosophy, law, sociology, political science, ethics, and more. Universities are among the few institutions where scholars from these diverse disciplines work side by side, making them uniquely capable of tackling AI’s multifaceted challenges.

Standardization isn’t just about equations and code — it’s about values, norms, and human rights. Universities are already cultivating this broader understanding.

3. Education and Future Workforce Development

The institutions educating the next generation of AI practitioners have a stake in shaping ethical norms. Embedding standards into curricula helps ensure that future engineers and researchers don’t just write better code — they build better systems.

Imagine if every computer science student graduated with a deep understanding of AI ethics, safety, and societal impact. Standards would quickly become embedded in developer culture.

4. Research Infrastructure and Long-Term Thinking

Unlike startups or corporations focused on quarterly results, university research labs are structured for longer time horizons and fundamental inquiry. They ask the questions others overlook: How should AI augment human decision-making? What does fairness mean across cultures? How should we structure global governance?

This long game perspective is critical when working on standards that might shape decades of technology.

5. Global Collaboration Networks

Universities already collaborate internationally through research partnerships, conferences, and exchange programs. These networks can be leveraged to coordinate standards work across borders — a necessity for a technology that doesn’t respect national boundaries.

What a University-Led AI Standards Initiative Could Look Like

To make this vision concrete, here’s one plausible roadmap for a university-led AI standards initiative:

1. Establish a Consortium of Diverse Institutions

This consortium would include universities from different countries and disciplines, ensuring a rich diversity of perspectives. It would be modeled on the W3C but tailored for AI, with working groups focused on specific domains (e.g., fairness, safety, privacy, explainability).

2. Partner with Industry and Government

Rather than working in isolation, the consortium would maintain formal partnerships with companies, governmental bodies, and nonprofit groups. This ensures that standards are both technically relevant and politically grounded.

Industry participation would be voluntary but incentivized — for example, companies could gain reputational benefits or fast-track access to research.

3. Publish Open Standards and Guidelines

All standards would be open, freely accessible, and developed through transparent processes. Multistakeholder input would be solicited through public comment periods, workshops, and iterative drafts.

4. Build Tooling and Compliance Ecosystems

Standards are only useful if they are actionable. Research departments would develop open-source tools for compliance, testing, and certification — similar to how web browsers adopted HTML/CSS validators.

5. Influence Policy and Regulation

The consortium would serve as a resource for policymakers, advising on legislation and best practices. This bridges the gap between academic research and real-world governance.

6. Educate the Next Generation

AI standards would be integrated into degree programs, professional certifications, and continuing education — ensuring that practitioners understand and adopt them as part of everyday practice.

The Benefits: Why This Matters

If done right, AI standards could:

• Make AI Safer for Everyone

Clear guidelines reduce the risk of harmful outcomes — whether that’s a biased hiring algorithm or a malfunctioning autonomous vehicle.

• Promote Fairness and Human Rights

Standards can embed values like fairness, accountability, and respect for privacy into AI development — helping to align technology with human dignity.

• Boost Innovation and Competition

Just as internet standards unlocked a global ecosystem of apps and services, AI standards can lower barriers, enabling new players to build on shared foundations.

• Enable Global Coordination

AI knows no borders. Standards facilitate cooperation between nations, reducing fragmentation and geopolitical tensions around technology.

• Increase Public Trust

People are more likely to embrace AI if they know it’s held to consistent, transparent standards developed through open processes.

• Prepare a Responsible Workforce

Engineers trained with ethical and technical standards at the center of their education will build better, safer systems from the start.

Addressing Common Concerns

“Standards will slow down innovation.”

On the contrary: clear baselines help innovators leap forward without constantly reinventing coordination puzzles.

“Universities are too slow or bureaucratic.”

Standards require care and deliberation. Universities are better suited for this than for-profit entities driven by quarterly results.

“Governments should lead this.”

Governments must play a role, but they lack the interdisciplinary expertise and technical depth universities offer. A collaborative model is the best path forward.

Conclusion

We live in a moment where artificial intelligence is not just changing technology, but reshaping society. To ensure this transformation benefits everyone — not just a select few — we need standards and guidelines that reflect shared values, rigorous science, and ethical foresight.

The World Wide Web became a global force for innovation because visionaries came together to build standards that everyone could trust. Now it’s time for the next generation of visionaries — particularly those in university AI research departments — to take the lead once again.

Universities bring the independence, depth, and public orientation necessary to craft AI standards that are not only technically sound, but socially responsible, globally relevant, and enduring. If we get this right, we won’t just build better AI — we’ll build a better future.