AI Safety in Focus: Quinnipiac Professor on Why Artificial Intelligence Risks Are No Longer Theoretical

A recent WFSB conversation between host Roger and Quinnipiac University Professor of Computer Science Chetan Jaiswal put a pressing question in plain terms: are warnings about artificial intelligence threatening humankind something the public should take seriously?

The segment, focused on AI safety concerns, comes as generative tools, autonomous systems and machine-learning models move from research settings into everyday life. Jaiswal, whose background is in computer science education and technical systems rather than policy-making, offered an expert perspective on how to separate realistic risks from science-fiction scenarios.

What AI safety actually covers

In public debate, the phrase AI safety can blur together several different problems. Jaiswal’s discussion highlighted that the risks are not one single threat but a spectrum. The more immediate issues include:

• Misuse by humans, such as fraud, disinformation campaigns, automated cyberattacks, or creation of deepfakes.
• Harmful or biased outputs generated by models trained on flawed data.
• Loss of control in complex automated systems where humans cannot intervene quickly enough.
• Privacy erosion through large-scale data collection and surveillance.
• Longer-term existential questions about machines that could become harder to govern as capabilities grow.

Distinguishing those categories matters. An AI system that produces a biased hiring recommendation is a real, documented problem today; a superintelligent system that evades human control is still a speculative scenario. Jaiswal’s framing emphasized that both deserve attention, but they require different kinds of safeguards, research and regulation.

Immediate risks versus long-term fears

One of the central points in the interview was the need to avoid treating all AI warnings as equal. Immediate safety issues are already measurable. Researchers and regulators have documented cases of automated decision tools producing inequitable results, language models fabricating information, and generative media making it harder for audiences to know what is real.

The conversation stressed that practical harms are no longer hypothetical — they are occurring across hiring, lending, education and media.

Longer-term fears about artificial intelligence threatening humankind, including the possibility that advanced systems might become misaligned with human values or too powerful to control, remain the subject of active academic debate. Jaiswal’s perspective as a computer science professor highlights that these concerns are taken seriously in technical communities, even if experts disagree on how probable or urgent they are.

The role of academic expertise

As a professor of computer science, Jaiswal represents the technical side of the AI safety conversation. His perspective is grounded in how models are built, trained and deployed, rather than in legislative or regulatory authority. That distinction matters because AI safety involves both engineering problems and political choices. An academic can explain why certain risks emerge from model design, but the decisions about acceptable risk often belong to broader society.

This is one reason the public conversation can feel confusing. Technical experts frequently disagree about the likelihood of catastrophic outcomes. Some argue that current systems are too narrow to pose existential risk; others contend that rapid increases in capability call for caution well before systems reach human-level intelligence. Jaiswal’s interview illustrated that these disagreements are not evidence that the field is ignoring safety — they are part of an ongoing effort to define it.

Why the debate is intensifying now

The WFSB segment arrives during a period of rapid adoption. Businesses, schools and government agencies are integrating AI tools faster than many oversight frameworks can keep up. That speed has made AI safety a live public issue, not an abstract academic question. Policymakers and technical organizations are working on guardrails, including risk management guidance and international policy cooperation.

For readers who want to examine the technical foundations of those safeguards, the NIST AI Risk Management Framework is a useful starting point. Stanford University’s Human-Centered AI initiative also publishes research on responsible AI and safety, and the OECD AI Policy Observatory tracks how different countries are approaching AI governance and risk.

What the public should watch

Rather than treating AI as either an unmanageable threat or a harmless novelty, Jaiswal’s comments suggested a middle path: pay attention to how systems are used, who is accountable when they fail, and what safeguards exist before deployment. For everyday users, that can mean checking whether an AI-generated claim is supported by other sources, being cautious about sharing personal information with chatbots, and recognizing that realistic images or audio may be synthetic.

The interview also served as a reminder that safety discussions should include a broad range of voices. Computer scientists can identify technical vulnerabilities, but teachers, health care workers, civil rights advocates and ordinary citizens are often the ones who encounter AI harms first. AI safety, in that sense, is not only about preventing a future catastrophe; it is also about managing existing risks in ways that are fair and transparent.

The WFSB conversation did not point to one catastrophic event. Instead, it was a reminder that the real AI safety discussion is already unfolding in schools, workplaces and public institutions — and that expert voices like Jaiswal’s are helping the public see the difference between responsible caution and alarmism.