B1 · Introduction to Artificial Intelligence

Strengths of Humans and Strengths of AI

AI's strengths are speed, scale, consistency and tireless attention to patterns in data. Human strengths are common sense, understanding of context, empathy, creativity, moral judgement and the ability to cope with the completely unexpected.

8 min Beginner Jump to the activity

What you need: Nothing. This page starts from the beginning, with no programming and no formulas.

Definition

Who would win at sorting a million photos? Who would win at comforting a sad friend?

AI's strengths are speed, scale, consistency and tireless attention to patterns in data. Human strengths are common sense, understanding of context, empathy, creativity, moral judgement and the ability to cope with the completely unexpected.

In simple words

A computer can check every payment in a bank in a second and never get bored. But it cannot tell that a customer sounds frightened on the phone, or decide what is fair when the rules don't fit. People get tired and make slips, but they understand what matters and why.

Why it matters

Knowing who is better at what helps decide which tasks to give to AI, which to keep for people, and which need both. This is how good teams, schools, hospitals and companies use AI.

Key terms

Scale
Doing a task for a huge number of cases, such as checking every payment in a country.
Consistency
Giving the same quality of result every time, without tiring or changing mood.
Empathy
Understanding and sharing what another person is feeling.
Judgement
Weighing many factors, including values and fairness, to make a good decision when rules do not settle it.
Moravec's paradox
The observation that tasks hard for people, like chess, are often easy for computers, while tasks easy for people, like walking, are hard for machines.

Explained in detail

What machines are built for

AI systems are strongest where a task is clearly defined, has lots of data, and must be done fast or at huge scale. A model can check millions of payments for fraud as they happen, scan every frame of a cricket match for the ball, or compare a medical image with thousands of past cases in seconds. It does not get bored at the ten-thousandth case, and it applies the same standard to the first and the last.

What people are built for

People are strongest where the situation is new, messy or human. We bring common sense (a wet floor is slippery), understanding of context (a joke at a party is not an insult), empathy (this patient is frightened, not rude), creativity (an idea that has never existed) and moral judgement (what is fair here?). We can also take responsibility, explain our reasons and be held to account. These are exactly the abilities that current AI lacks or only imitates.

The surprise: easy and hard are reversed

In the 1980s the robotics researcher Hans Moravec pointed out something strange. Computers found it relatively easy to play chess or solve algebra, which people find hard, but extremely hard to walk across a room or pick up a cup, which a toddler can do. Abilities like seeing, moving and handling objects were shaped over millions of years of evolution and feel effortless to us, so we underestimate how complex they are. This is why a chess program beat the world champion in 1997, while robots that can tidy a messy room are still being developed.

Together is better than either

After the chess computer Deep Blue defeated Garry Kasparov in 1997, players began experimenting with teams of a person and a computer working together, sometimes called "centaur" or freestyle chess. In tournaments in the mid-2000s, such teams, even with average players, could beat both top grandmasters and strong computers playing alone. The people chose strategy and judged unusual positions; the computers checked moves deeply. The lesson reaches far beyond chess: the best results usually come from combining what machines and people each do well.

SituationBest handled byWhy
Reading 50,000 product reviews todayAIHuge volume, clear task
Telling a family that a loan was refusedA personNeeds empathy and explanation
Screening X-rays, then deciding treatmentBoth togetherAI flags; the doctor decides

How it works

AI is usually stronger atPeople are usually stronger at
Speed: millions of checks per secondCommon sense about everyday life
Scale: huge amounts of dataEmpathy: understanding feelings
Consistency: same answer every time, no tirednessCreativity: truly new ideas and purposes
Spotting small patterns, such as early signs in a scanMoral judgement: what is fair and right
Working all night without a breakHandling a situation nobody expected

Try it yourself

Who is better at what?

Watch a hospital's day pass between the two.

Human, AI or a team?

Who should do each job? Think about speed, scale, feelings and fairness.

  • Checking ten million photos for duplicates

  • Deciding a fair punishment in a court

  • Reading X-rays in a busy hospital

  • Translating thousands of road signs overnight

  • Writing the story for a new film

  • Telling a family some difficult news

Common mistakes

  • Thinking AI is better at everything because it is fast. Speed is not wisdom.
  • Thinking people are never needed once AI arrives. AI changes tasks; people stay responsible.
  • Thinking AI is never wrong because it is consistent. It can repeat the same mistake millions of times.

Check yourself

Question 1

1. Which is a human strength?

Question 2

2. Why is a team of AI and a doctor good for reading scans?

Question 3

3. What is a danger of AI's consistency?

Summary

  • AI is strong at speed, scale, consistency and spotting patterns.
  • People are strong at common sense, empathy, creativity and moral judgement.
  • Many of the best results come from people and AI working together.

Finished reading?

Sources: Elements of AI (University of Helsinki)