GoalWeigh the benefits and drawbacks of automation and AI for people, work and society, and give balanced reasons for a decision.
New words
productivity · how much work or how many products are finished in a given timeretrain · learn new skills so that a person can do a different jobbias · a result that is unfair to some people, often because the training data was one-sidedoversight · people checking and controlling what a system does, especially when its decisions matter
Explanation
Automation and AI bring benefits and drawbacks, and a fair judgement looks at both. A machine can work through the night, repeat a task in exactly the same way and do dangerous jobs, so people are safer and less material may be wasted. But machines are expensive to buy and maintain, and a machine is only as good as its program: it may fail on a case nobody planned for. Some tasks that people did are now done by machines, so jobs change: some shrink and new ones appear, such as maintaining, programming and checking, which may need retraining. Nobody can say exactly how jobs will change, so be careful with predictions. AI also raises questions of bias and privacy, so people need oversight of important decisions.
Benefit
Drawback
Question to ask
works through the night
high cost to buy and run
does it pay for itself?
same result every time
only as good as its program
what if something unexpected happens?
can do dangerous work
some tasks move from people to machines
who will do the new jobs, and who needs training?
Worked examples
Year
Saved so far
Still to recover
1
18 000
72 000
2
36 000
54 000
3
54 000
36 000
4
72 000
18 000
5
90 000
0
Does a machine pay for itself? (invented numbers) A sorting machine costs 90 000 to install. Each year it saves 24 000 but costs 6 000 to run, so the net saving is 18 000 a year. After 5 years the savings equal the cost, so the machine must work for 5 years or more to pay for itself.
Task
Choice
Reason
check 2 000 identical screws for flaws
machine
repetitive and precise
comfort a nervous patient
person
needs care and understanding
plan a school garden
both
machine does sums, people decide
Person, machine or both? Think about the task, then give a reason. Checking thousands of identical screws is repetitive and precise, so a machine suits it. Comforting a nervous patient needs care, so a person suits it. Planning a school garden suits both: a machine can work out costs, but people decide what they want.
Job
Before
After
sorters
40
24
machine technicians
0
8
data checkers
0
5
programmers
0
3
total
40
40
Jobs change. In an invented parcel depot, 40 people sorted parcels by hand. After robots arrive, there are still 40 jobs, but of different kinds. Real cases differ, and people may need training to move to the new roles.
Class activity
Machine, person or both (no computer). Write these eight tasks on slips of paper: count the stock in a store every night; teach a shy child to read; paint the same logo on 1 000 boxes; decide who is telling the truth in an argument; clean the inside of a chemical tank; write a class poem; sort recycling by material with a camera; give first aid to a hurt pupil. In pairs, sort the slips into three piles: machine, person, both. Write a reason on each slip. Swap piles with another pair and challenge one decision. Finish with one sentence about why a person should check what an AI tool produces.
Practice
1
A machine costs 48 000 to install. It saves 14 000 a year and costs 2 000 a year to run. After how many years has it paid for itself?
The net saving is 14 000 − 2 000 = 12 000 a year. 48 000 ÷ 12 000 = 4 years.
A school uses an AI tool to suggest which pupils need extra help. It was trained only on last year’s results of one class. Name one risk and two things the school should do.
Risk: the suggestions may be unfair or wrong for pupils unlike that class. The school should check the suggestions with teachers (oversight) and add more varied data and test the tool again.
4
One person says: “Automation always takes more jobs than it makes.” Another says: “Automation always makes more jobs than it takes.” Why are both statements too simple?
Different cases give different results. Some jobs shrink, others appear, and it depends on the industry and on training. Nobody can say that one result is true for every case.