Routes How much will AI add to economic growth, and why do the estimates differ so much?
How much will AI add to economic growth, and why do the estimates differ so much?
Some economics helps. A spreadsheet and multiplication are all the arithmetic needed.
kosplora-route/0.2 · by Bentley Moon
What would change this route
If a measurement of AI's effect on productivity across a whole economy were published, station 5 would set both estimates against it.
The whole route, before you start
- 1 · orient · 20mRead the paper's summary and introduction.
- 2 · source · 1hRead section 3 of the paper, where the estimate is built.
- 3 · practice · 40mRebuild the estimate in a spreadsheet, one multiplication to a cell, and enter your figures in the check below.
- 4 · counterpoint · 40mRead a reply that keeps the same formula and changes the inputs, and arrives at a gain about ten times larger.
- 5 · make · 40mMake a one-page table: the four inputs from each paper, the result of swapping in each of the reply's inputs one at a time, and which single disagreement moves the answer most.
- 6 · person · 15mShow the table to someone who studies economics and ask which input they find least believable in each paper.
- 7 · reflect · 10mRecord which estimate you find more believable now, and why.
Optional study preferences. These stay in this browser and do not change the route’s steps.
Station 1 · orient · 20m
Read the paper's summary and introduction. Write down its estimate for productivity over ten years, its estimate for output, and why it calls them modest.
The Simple Macroeconomics of AI · Daron Acemoglu · 2024-05
Station 2 · source · 1h
Read section 3 of the paper, where the estimate is built. Write down its four inputs: the share of tasks exposed to AI, the share of those profitable to automate within ten years, the saving in labour costs, and the labour share it multiplies that saving by. Beside each, note the study it comes from.
Station 3 · practice · 40m
Rebuild the estimate in a spreadsheet, one multiplication to a cell, and enter your figures in the check below. Where your figure differs in the last place from the printed one, find whether the paper rounded before or after multiplying.
Station 4 · counterpoint · 40m
Read a reply that keeps the same formula and changes the inputs, and arrives at a gain about ten times larger. List each input it changes and the reason it gives. Recompute its product yourself and check the last step of its equation against its stated yearly figure.
AI and Growth: Where Do We Stand? · Philippe Aghion and Simon Bunel · 2024-06
Station 5 · make · 40m
Make a one-page table: the four inputs from each paper, the result of swapping in each of the reply's inputs one at a time, and which single disagreement moves the answer most. End with one sentence on what evidence would settle that input.
Station 6 · person · 15m
Show the table to someone who studies economics and ask which input they find least believable in each paper. Write their answer beside yours.
Station 7 · reflect · 10m
Record which estimate you find more believable now, and why. Both papers leave out AI that speeds up research itself; if you want to see what that does, the interactive model below lets you change it and watch the economy respond.
GATE: An Integrated Assessment Model for AI Automation · Ege Erdil, Andrei Potlogea, Tamay Besiroglu, Edu Roldan, Anson Ho, Jaime Sevilla and others · 2025-03-06
What you leave with
A one-page table setting two published estimates of AI's effect on productivity side by side, input by input, with your own recomputation, the input that moves the answer most and an economist's objection.
The standard it is checked against
ACM Artifact Review and Badging, version 1.1: Results Reproduced
Association for Computing Machinery · 2020-08-24
Someone other than the author obtains the central result from the author's own inputs. Here the result is Daron Acemoglu's estimate that AI will raise total factor productivity by about 0.66% over ten years, and the inputs are the four numbers his paper prints.
Who checks it: You, against the figures printed in the paper, and then one other person who recomputes one figure from the same four inputs.
Kosplora does not issue a certificate. A partner who recognises this standard would assess the report; no partner has agreed to yet.
When you are done
- study preferences, completed steps and notes in one file
- Download the route file edit a copy in any text editor
- Read the route file format
- Or close the page.