What AI Can and Cannot Do

Reality check: AI is astonishing at some things and surprisingly weak at others. Knowing the difference helps you use it well and avoid trusting it blindly.

What AI is great at

• Spotting patterns in huge amounts of data
• Recognising images, speech, and text
• Generating text, images, and code
• Doing repetitive tasks tirelessly and fast
• Personalising recommendations

What AI struggles with

• True understanding — it predicts patterns, it does not "know" things the way you do
• Common sense and reasoning about the real world
• Being reliably correct — it can state false things confidently (called hallucination)
• Anything outside its training — new or rare situations
• Genuine creativity, ethics, and judgement

A useful rule of thumb: AI is excellent at tasks that are narrow, repetitive, and data-rich, and weak at tasks needing broad context, judgement, or guaranteed accuracy.

💡 Takeaway: treat AI as a brilliant, fast, but sometimes-wrong assistant — never as an infallible authority.