AI's Broader Impact on Society, Policy, and Global Governance

UPI Study › Courses › Introduction to Artificial Intelligence › Free lessons › Lesson 9.8 Free with a free account

AI's Broader Impact on Society, Policy, and Global Governance

Introduction to Artificial Intelligence · Chapter 9: AI Applications, Ethics, and the FutureSection 9.8 · lesson 69 of 6915 min · free with a free account

← The Evolving Workplace: AI's Influence on Jobs and Skills

Earn 3 college credits for Introduction to Artificial IntelligenceRecommended for credit by NCCRS · Computer Science 373 · self-paced · one payment, lifetime access
Enroll in Introduction to Artificial Intelligence · $250 →Download the syllabus (PDF)

In this lesson

How AI affects information, surveillance and access, and how nations govern it.

  • Explain how deepfakes and sharing rates drive misinformation, and why base rates make surveillance alerts mostly false
  • Describe the digital divide and AI's effects on education, democracy and culture
  • Compare national approaches to AI regulation and the risk tiers of the EU AI Act
  • Identify international agreements and oversight tools, such as independent audits and the four-fifths rule

Free study notes

Get this lesson's notes by email

The key points and 4 practice questions with answers, in your inbox. One email, free.

One email. Unsubscribe in one click.

Transcript

Everything said in the video, section by section (1,840 words). Read along, or search it.

Opening Question

Face Recognition in Every Transit Station

Mei sits on a city council. Next week she votes on a plan to put AI face recognition in every transit station, to find wanted people. The vendor says it's 99 percent accurate. Should she vote yes? And who decides the rules for tools like this, and can those rules keep pace?

Looking Outward

Society, Institutions and Changing Rules

Earlier lessons covered the ethics of responsible AI. This one looks outward: at what a society sees, how it's watched, who gets opportunity, and how its institutions work. Societies respond with laws, treaties, standards and oversight bodies. These rules change fast, so the legal details here are as of October 2026. Check official sources before relying on them.

Misinformation and Deepfakes

Cheap Fakes and the Liar's Dividend

Misinformation is false or misleading content that spreads, whatever the sharer intends. A deepfake is realistic audio, video or an image, made or altered by AI, showing something that never happened. Generative AI makes fakes cheap, fast and personal, tailored to one target. In a widely reported 2024 case, an employee in Hong Kong transferred about 25 million U.S. dollars, after a video call with deepfake versions of colleagues. A subtler harm is the liar's dividend. Once people know fakes are possible, anyone caught on a genuine recording can claim it was fake, and trust in all evidence falls.

Spread Model

A Sharing Rate of One Is the Pivot

Why does false content spread so far? A simple model: each person who sees a post passes it to r new people, on average. It ignores overlap and much else. At r of 1.3, one post reaches about 13 people in five rounds of sharing, and 56 in ten. It grows without limit. At r of 1, it grows steadily: 11 people in ten rounds. At r of 0.8, it fades out, reaching only about 4.6. So the pivot is r equals one. Friction, like labels, prompts to read before sharing, and limits on mass resharing, lowers r. Cutting 1.3 to 0.8 removes about 92 percent of the reach. You don't need to stop every false post, only push the average below one.

Surveillance

Watching People at Scale

AI also makes it practical to watch people at scale: face recognition in public places, tracking movements and purchases, analyzing communications. Used carefully, it can find missing persons or catch criminals. Used carelessly, it can chill free speech and assembly, enable discrimination, and concentrate unaccountable power.

Base-Rate Trap

Most Watch-List Alerts Are False

And there's a statistical trap. A system scans 100,000 people to find 10 on a watch list, and catches 99 percent of them: 9.9 real matches. With a 1 percent false alert rate on everyone else, that's about 1,000 false alerts. Only about 1 percent of alerts are real. Even at 0.01 percent, half the alerts are false. That's the base-rate effect, and it's why rules often restrict public biometric identification, and require a human to check every match.

Digital Divide

Who Is Online and Who Is Not

AI's benefits depend on being connected. The digital divide is the gap between people with meaningful access to technology and those without it. The International Telecommunication Union's 2025 figures: about 6 billion people are online, and about 2.2 billion are not. Ninety-six percent of those offline live in low- and middle-income countries. In high-income countries, 94 percent are online. In low-income countries, 23 percent. Using AI well also takes fast connections, devices, skills, and content in your own language. Without deliberate effort, its gains flow mainly to the already connected.

Quick Check

Real Alerts at a Stadium

Quick check. A stadium system scans a million fans to find 50 people on a list. It catches 90 percent of them, and wrongly flags 0.01 percent of everyone else. Roughly what share of its alerts are real? Pause, and work it out. About one in three. It finds 45 of the 50. And 0.01 percent of nearly a million other fans is about 100 false alerts. So 45 of about 145 alerts are real: 31 percent. Nine in ten is the trap. That's the catch rate, not what an alert means.

Institutions

Education, Democracy and Culture

AI reaches whole institutions. In education, adaptive tutors and help for learners with disabilities are promising. But a tool can do the work for the learner. The question is whether it builds skill or replaces it. Democracy depends on shared facts and trust. Fakes, automated accounts imitating crowds, and messages tailored to each voter's fears strain both. Responses include labeling AI content, political ad transparency and fast fact-checking. In culture, AI gives creators new tools and helps preserve languages. But it works well in only a small fraction of the world's roughly 7,000 languages, and whether training on creative work needs permission and pay is contested.

National Approaches

EU, United States, China and United Kingdom

Countries regulate differently. The European Union has one comprehensive, risk-based law, the AI Act, in force since 2024, with duties phasing in. The United States has no single federal AI law. Federal policy favors light regulation, and a December 2025 executive order directs federal challenges to state AI laws. But state laws, such as California's, remain in force, and Congress hasn't passed a replacement. China uses targeted rules for specific uses, with labeling of AI-generated content in force since September 2025. The United Kingdom has existing regulators apply five principles, with no standalone AI law.

EU AI Act Tiers

Banned, High, Limited and Minimal Risk

The EU AI Act rests on proportionality: the higher the potential harm, the stricter the rules. Unacceptable-risk uses, like social scoring, are banned. Amendments in 2026 added a ban on AI that generates non-consensual intimate images, and child sexual abuse material, applying from December 2026. High-risk systems, like AI used in hiring or credit, face duties: risk management, good data, documentation, logging, human oversight and testing. Limited-risk systems have transparency duties: people must know they're talking to a machine, and AI content must be marked. Minimal-risk systems, like spam filters, get no new duties. General-purpose model providers have their own duties, like a copyright policy and a training-content summary, with more for models posing systemic risk.

Timeline

When the AI Act's Duties Apply

The rules phase in. The Act entered into force on August 1, 2024. Bans and an AI literacy duty applied from February 2, 2025, and general-purpose model duties from August 2, 2025. A July 2026 amendment package, the Digital Omnibus, changed the schedule. Most transparency duties apply from August 2026. Systems already on the market have until December 2, 2026, to mark AI content. National sandboxes are due by August 2, 2027. Stand-alone high-risk duties moved to December 2, 2027, and high-risk AI in regulated products to August 2, 2028. Standards and authorities weren't ready. It's a reprieve, not a repeal.

Penalties

Fines in Three Tiers

Fines come in tiers. For banned practices, up to 35 million euros or 7 percent of worldwide annual turnover, whichever is higher. For most other duties, 15 million or 3 percent. For misleading information to authorities, 7.5 million or 1 percent. For small and medium-sized enterprises, including start-ups, it flips to whichever is lower. So a large company, with 2 billion euros of turnover, faces up to 140 million for a banned practice. Above 500 million in turnover, the percentage takes over. And these are maximums.

Quick Check

Maximum Fine for a Start-Up

Quick check. A start-up that qualifies as a small or medium-sized enterprise has worldwide turnover of 60 million euros. It commits a banned practice. What's the maximum fine? Pause, and work it out. Four point two million euros. Seven percent of 60 million is 4.2 million, and for an SME, the lower of that and 35 million applies. Thirty-five million is the trap: that's the rule for large firms. And 1.8 million uses the 3 percent tier, which is for other duties.

Global Cooperation

Principles, Declarations and a Binding Treaty

AI crosses borders. The OECD AI Principles of 2019 set shared values, and UNESCO's 2021 Recommendation on AI ethics was adopted by all its member states. The 2023 Bletchley Declaration brought governments together on the safety of advanced AI, followed by later summits. The Council of Europe's Framework Convention, opened in 2024, is the first legally binding international AI treaty. The EU ratified it in May 2026. It takes effect after five ratifications, including three Council of Europe members. And at the UN, an independent scientific panel assesses AI, and a Global Dialogue lets every member discuss governance. The first met in Geneva in July 2026, and the next is planned for New York in May 2027. Countries differ in values and interests, so recent efforts stress interoperability: national systems that can work together, not one global rulebook.

Oversight Tools

Audits, Red-Teaming and Standards

Laws matter only if someone checks they're followed. Oversight runs from internal reviews and impact assessments, to independent audits by outside parties. Model evaluation and red-teaming try to make a system fail before release. Incident reporting shares serious failures. Regulatory sandboxes let new AI be tested with a regulator's guidance. And standards set benchmarks for good practice, like ISO 42001 for AI management systems, and the NIST AI Risk Management Framework.

Audit Check

The Four-Fifths Rule

One concrete audit check is the four-fifths rule, from U.S. employment guidance. If a group's selection rate is below 80 percent of the highest group's rate, the gap is flagged. In the loan example from the ethics lesson, one group was approved at 0.56, and the other at 0.44. The ratio is 0.44 over 0.56: 0.786, below 0.8, so it's flagged. Equalize approval rates, and the ratio becomes one. But an auditor would check the other fairness measures too, and a flag prompts investigation. It isn't proof of unlawful bias.

Long-Term Challenges

Pacing, Systemic Risk and Legitimacy

Some challenges outlast any single law. The pacing problem: technology changes faster than legislation. Responses include rules based on risk, not specific technologies, regular review, and sandboxes. Advanced general-purpose models create hard-to-predict risks. The EU presumes systemic risk for models trained with more than 10 to the 25 operations. Fairness between countries: if a few set the rules and own the infrastructure, others live with choices they didn't make. And legitimacy: transparency, public input, independent scrutiny and redress. No single tool is enough. Good governance layers laws, standards, audits and public scrutiny, and keeps adjusting.

Back to Mei

Questions Before the Vote

Back to Mei's vote. The vendor's 99 percent hides the base-rate trap: when wanted people are rare, most alerts point at innocent riders. So she asks how many alerts would be real, who checks each match, and what rules apply. In the EU, real-time face identification in public spaces is restricted under the AI Act's banned practices. And she can require oversight: an independent audit, results by group, incident reporting, and a public review before any expansion.

Key Takeaways

AI and Society in Summary

To sum up. AI makes fakes cheap, fast and personal, and spread depends on whether r is above one. Rare targets make surveillance alerts mostly false. The digital divide means AI's gains flow first to the connected: 94 percent online in high-income countries, 23 in low-income ones. The EU uses one risk-based law, with fines up to 35 million euros or 7 percent of turnover. The U.S. relies on a patchwork, China on targeted rules, and the U.K. on regulators. Cooperation runs from principles to a binding treaty and a UN dialogue, and oversight layers audits, evaluations, reporting and standards.

Think About It

A Shocking Video Before an Election

Here's something to think about. Two days before an election, a realistic video shows a candidate saying something shocking. What should platforms, journalists and voters do in those two days? Which tools from this lesson would help, and what could the liar's dividend do?

About Introduction to Artificial Intelligence at UPI Study

Introduction to Artificial Intelligence (Computer Science 373) is a self-paced online course of 69 video lessons in 9 chapters, from intelligent agents and search to game strategy, neural networks, logic and reasoning, machine learning, language, robotics and AI ethics. It carries a NCCRS credit recommendation for 3 semester credits.

Every lesson transcript is free on pages like this one, and the videos are free with a free UPI Study account. Enrolling adds a practice quiz after each lesson, the graded work and final exam that earn the credit, lifetime access, and an official transcript on completion. UPI Study has helped over 130,000 students with college credits. ACE and NCCRS have over 2,100 cooperating colleges that will consider accepting credits; the final decision is with the universities.

Enroll now → Course details Download the syllabus All Computer Science courses Every course with PRO →

Compare: AI vs Current Trends in CS and IT

Why take it here

Why students take Introduction to Artificial Intelligence with UPI Study

How machines search, reason and learnIntroduction to Artificial Intelligence covers how AI systems are built: intelligent agents, search and heuristics, game strategy, logic, neural networks, machine learning, language and perception, and the ethics of AI. It is a common upper-level course in computer science degrees.
Credit at a fraction of the priceThe same 3 credits can cost $750 or more at a US college. Here it is one payment, self-paced, with lifetime access.
Finish your degree soonerTransfer credits reduce the courses you still owe, so you can graduate earlier or lighten a heavy semester.
Strengthen an applicationA completed college-level course on an official transcript shows admissions and scholarship committees you can do the work.
Study on your own scheduleNo start dates, no deadlines. Watch the lessons, practice, and take the final when you are ready.
130,000+students helped with college creditsACE and NCCRS have over 2,100 cooperating colleges that will consider accepting credits; the final decision is with the universities.

Where the credit goes

Universities that work with NCCRS

Introduction to Artificial Intelligence carries a NCCRS credit recommendation. These universities work with NCCRS; some are UPI Study's signed partners. ACE and NCCRS have over 2,100 cooperating colleges that will consider accepting credits; the final decision is with the universities.

Thomas Edison State UniversityThomas Edison State UniversitySigned partnerUniversity of the PeopleUniversity of the PeopleSigned partnerCharter Oak State CollegeCharter Oak State CollegeSigned partnerStrayer UniversityStrayer UniversitySigned partnerColumbia Southern UniversityColumbia Southern UniversitySigned partnerIndiana TechIndiana TechSigned partnerPenn StatePenn StateWorks with NCCRSSouthern New Hampshire UniversitySouthern New Hampshire UniversityWorks with NCCRSSUNYSUNYWorks with NCCRS +many more cooperating universitiesFind my college →

Find my college → Not listed? Ask Athena in the corner of this page, or book a 15-minute call.

FAQ

Questions students ask us

About AI's Broader Impact on Society, Policy, and Global Governance

What is a deepfake, and what is the liar's dividend?
A deepfake is realistic audio, video or an image made or altered by AI to show something that never happened. Generative AI makes such fakes cheap, fast and tailored to one target. The liar's dividend is a subtler harm: once people know fakes are possible, anyone caught on a genuine recording can claim it was fake, so trust in all evidence falls. Labels and fast fact-checking are common responses.
Why do face recognition systems in public places produce so many false alerts?
When wanted people are rare, false alerts on the huge innocent population can far outnumber true matches, which is the base-rate effect. Scanning 100,000 people for 10 on a watch list with a 99 percent catch rate gives 9.9 real matches, but a 1 percent false alert rate adds about 1,000 false alerts, so only about 1 percent of alerts are real. That is why rules often require a human to check every match.
How does the EU AI Act classify AI systems by risk?
The EU AI Act sorts AI uses into tiers, with stricter rules for higher potential harm. Unacceptable-risk uses, such as social scoring, are banned. High-risk systems, such as AI used in hiring or credit, face duties including risk management, good data, documentation, logging, human oversight and testing. Limited-risk systems have transparency duties, minimal-risk systems such as spam filters get no new duties, and general-purpose model providers have separate duties.
What is the four-fifths rule in AI fairness audits?
The four-fifths rule, from U.S. employment guidance, flags a possible disparity when a group's selection rate is below 80 percent of the highest group's rate. If one group is approved at 0.56 and another at 0.44, the ratio is 0.44 divided by 0.56, or about 0.786, which is below 0.8, so the gap is flagged. A flag prompts investigation and is not, on its own, proof of unlawful bias.

About the Introduction to Artificial Intelligence course

How many college credits is Introduction to Artificial Intelligence?
Introduction to Artificial Intelligence (Computer Science 373) carries a NCCRS credit recommendation for 3 semester credits. ACE and NCCRS have over 2,100 cooperating colleges that will consider accepting credits; the final decision is with the universities.
How much does Introduction to Artificial Intelligence cost and what is included?
$250, paid once, with lifetime access: all 69 video lessons, a practice quiz after each, the graded work and final exam, and an official transcript when you finish. Students who want several courses choose UPI Study PRO instead.
Which universities accept UPI Study credits?
UPI Study has signed partner agreements with universities such as Thomas Edison State University, University of the People and Charter Oak State College, and works with universities such as Penn State, SNHU and SUNY. ACE and NCCRS have over 2,100 cooperating colleges that will consider accepting credits; the final decision is with the universities. Use "Find my college" on this page to check yours.
How long does Introduction to Artificial Intelligence take?
It is self-paced with no deadlines. The 69 lessons total about 16 hours of video; most students add reading and practice time and finish in a few weeks to a few months, on their own schedule.
How are the exams done?
Quizzes and assignments are online in the course. The final exam is taken online with proctoring; a person on the UPI Study team checks assignments, usually within two days.
How do my credits reach my college?
When the course shows 100% complete, request the official transcript from your dashboard. It is sent to the college you name, and you can send it to more than one. Check your college first with "Find my college" on this page, or ask Athena.

Ready for the full course? Earn 3 credits in Introduction to Artificial Intelligence.

69 video lessons, a quiz after each one, and the graded work that earns the credit. Recommended for credit by NCCRS.

Enroll in Introduction to Artificial Intelligence →Download the syllabusEvery course with PRO →
Introduction to Artificial Intelligence
3 credits · NCCRSEnroll →