Question 1

What is a CAPTCHA and why are they getting harder?

Accepted Answer

**CAPTCHA** stands for **Completely Automated Public Turing test to tell Computers and Humans Apart**. It is a challenge designed to be easy for people and difficult for software.

**Why they are getting harder** is straightforward and slightly uncomfortable: **machines got better at them than humans**.

Optical character recognition surpassed human accuracy on distorted text years ago, which is why text CAPTCHAs became progressively more mangled until they were failing legitimate users, and were then largely abandoned. Image recognition then matched and exceeded human performance on object identification, undermining the "select all the traffic lights" generation. Each defence was defeated, and the response was to make the challenge harder — until the challenge is now frequently harder for people than for the bots it targets.

**How modern systems actually work.** The visible puzzle is increasingly not the real test. Systems such as reCAPTCHA v3 and similar services score requests **invisibly**, based on:

**Mouse movement and timing** — human cursor paths are irregular in characteristic ways.

**Typing cadence**, scroll behaviour and interaction patterns.

**Browser and device fingerprinting**, including inconsistencies suggesting automation.

**IP reputation** and the history associated with it.

**Cookies and prior browsing**, including whether you are signed in to the provider's other services.

The checkbox is largely theatre; the decision is usually made before you click it. You are shown a puzzle only when the score is ambiguous.

**Why you get harder challenges than other people.** Using a VPN, a privacy-focused browser, blocking cookies, a shared or datacentre IP address, or an unusual configuration all lower your score — so the more privacy measures you use, the more CAPTCHAs you face.

**The accessibility problem is serious.** Visual challenges exclude blind users, audio alternatives are frequently worse, and both are difficult for people with cognitive or motor impairments.

**Alternatives** include cryptographic attestation and proof-of-work challenges.

Question 2

What does digital accessibility actually require?

Accepted Answer

That people with disabilities can perceive, operate, understand and use a digital service — and the practical standard almost everywhere is WCAG, whose four principles are exactly those four words.

Question 3

How does facial recognition actually work?

Accepted Answer

By converting a face into a numerical vector and comparing distances between vectors — not by matching pictures, which is the misconception underlying most public discussion of it. The pipeline:

Question 4

What is an AI agent, and how is it different from a chatbot?

Accepted Answer

An agent pursues a goal over multiple steps, deciding what to do next and taking actions in the world; a chatbot responds to messages. The difference is autonomy and the ability to act rather than

Question 5

How does OCR work?

Accepted Answer

By locating text in an image and then recognising it — two separate problems, and the second is far easier than the first, which is why scanning a clean document works well and photographing a sign

Question 6

How are deepfakes made, and can they be detected?

Accepted Answer

By training a model to map one person's appearance or voice onto another's performance — and detection is possible but increasingly unreliable, which is why the response has shifted toward provenance

Question 7

How are AI models actually evaluated, and what do benchmarks mean?

Accepted Answer

Against standardised test sets with published scores — and the honest position is that benchmarks measure something real, measure it narrowly, and are increasingly unreliable as headline comparisons

Related Questions