AI and Society

Most of this track so far has been about AI in your pocket — the apps you use, the data they collect. This mission zooms out. AI isn’t only changing how you scroll; it’s changing how hospitals diagnose, how schools teach, how banks lend, how governments decide who gets what. The effects cut both ways. A well-designed AI system can catch a cancer earlier, route a food delivery faster, or translate a lecture into Thai in real time. A poorly designed one can deny a loan to the wrong group, push misinformation at scale, or leave people without recourse when it’s wrong. Understanding how AI reshapes society — the wins, the costs, and the choices nobody asked you about — is the last piece of AI literacy.

For most of the computer era, software was used by specialists — accountants ran spreadsheets, doctors read ultrasound machines, pilots flew autopilots. Software did narrow, pre-programmed tasks and nothing more. Industries used it, but they weren’t transformed by it.

That started to change in the 2010s, when machine learning quietly leaked into every field. Insurance companies used it to price risk. Hospitals used it to prioritize which chest X-rays a doctor should look at first. Banks used it to score loan applications. Schools used it to flag students who might drop out. Courts, in some countries, used it to estimate re-offense risk during sentencing. Most of these uses were invisible to the people affected by them.

The arrival of generative AI in 2022 made AI visible. Suddenly anyone could open ChatGPT and see AI write, explain, and code — and the conversation about AI moved from tech blogs to dinner tables. Offices, newsrooms, law firms, and schools started debating what AI should and shouldn’t do, often in real time, without much preparation.

That’s where we are now: AI is in every industry but hasn’t been absorbed by any of them yet. Rules are being written as we go. The decisions made in the next few years — by governments, by companies, by ordinary users — will shape what AI ends up being used for, and whose interests it serves. This mission is about stepping into that conversation, not just watching it happen to you.

Jobs will change, not disappear — but the change is uneven. Routine work that follows predictable patterns (scanning documents, drafting boilerplate emails, entry-level customer support) is the easiest for AI to handle. Creative, hands-on, or highly interpersonal work is slower to change. The pattern isn’t humans vs AI — it’s humans + AI vs humans alone, with the former usually winning. A radiologist with AI reads scans faster and catches more. A paralegal with AI drafts cleaner first passes. But people whose jobs were purely those routine tasks are forced to adapt, and adaptation isn’t always easy or fast.

Fairness is more than a philosophy question. AI systems trained on historical data reflect historical bias. A hiring tool trained on past hires of one demographic will prefer that demographic. A loan model trained on past approvals will repeat past exclusions. These are not bugs — they’re the system working as designed, reflecting the data it was shown. Fixing them requires changing the data, auditing the outputs, and sometimes choosing accuracy losses in exchange for fairer results. These are value choices, not technical ones.

Human oversight isn’t optional for decisions that matter. When AI recommends a movie and it’s wrong, you scroll past. When AI recommends a medical treatment and it’s wrong, someone could die. The higher the stakes, the more important it is to keep a human in the loop — reviewing, correcting, and taking responsibility for the final call. Regulators in the EU, the US, and elsewhere are starting to write rules that require this explicitly for high-risk AI. Thailand’s own AI policy conversation is still early.

Access is unequal. The best AI tools cost money, run on expensive hardware, and need fast internet. That means the gap between people who use AI well and people who don’t could become another version of the gap between people with good education and people without. Making AI accessible — through free tiers, public training, and schools that teach these skills — is how societies prevent AI from widening the gap instead of narrowing it.

You’ve reached the end of the Engaging with AI track. Here’s what you can now do that most people can’t:

Recognize AI when you see it, in apps and services you use every day. Understand how AI learns from data — and why bad data makes bad AI, no matter how clever the algorithm. Verify AI outputs instead of trusting them blindly. See what your apps collect about you and make deliberate trade-offs. Think critically about AI’s role in wider society — the jobs it changes, the fairness it tests, the oversight it needs.

From here you can continue to the next track: Creating with AI — where you learn to use AI as a tool yourself, starting with prompt engineering. Or, if you’d rather go deeper into how AI actually works under the hood, the Deep Dive track builds from math foundations up through modern language models.