Embracing Artificial Intelligence by Joshua Gans

Source details

Interview format: conversation between Joshua Gans and Marjorie Henriquez for F&D Magazine.
Publication date: September 2023.
Core argument: Assessing the power and pitfalls of AI requires putting it in the hands of people to learn its benefits and adverse consequences.

Key distinction:
- Previous automation focused on predominantly physical tasks; AI provides the potential to automate cognitive tasks.
- Past automation of physical tasks also included cognitive expression of intent; information technology automated computational cognitive processes.
Unique AI capability:
- Ability to interact in variable and non-standardized environments (for example, handling random objects not in particular locations), requiring a sense of the environment.
Implication:
- AI extends automation into more varied contexts than earlier technologies.

Observed effects where AI is deployed in workplace tasks:
- Tools often enable people with lower skills and less experience to perform at the level of those with higher skills and more experience.
- Example: demand prediction tools in Tokyo allowed less experienced taxi drivers to waste less time finding riders, with little productivity improvement for experienced drivers.
Projected outcomes (next five years):
- Skill premia in certain occupations will be reduced.
- Employment opportunities will open up for a wider set of people.
- Likely effects: increase productivity, raise employment, and lower income inequality—at least at certain ranges of the income distribution.
Longer-term forecast: Beyond the next five years, outcomes are harder to predict.

Rapid improvements in generative AI (LLMs such as ChatGPT or Bard, generative art, coding boosts) have led to extrapolations that people will be quickly replaced in affected tasks.
Gans’ assessment:
- Much of the public angst is overhyped and is vague speculation rather than evidence-based.
- Productivity improvements from AI are real and noticeable.
- Policy stance: “letting it rip” is the right policy for the current moment to capture productivity gains and to learn about adverse consequences.
Cautionary note:
- Rapid adoption necessitates active monitoring for adverse consequences, identifying causes, and experimenting with policy interventions to mitigate harms.

For organizations like the IMF:
- Standard playbook applies: AI should be as competitively provided as possible and available as widely as possible (through trade) so it can be deployed wherever it increases productivity.
- Mission alignment: similar to programs encouraging information technology and internet access.
On regulation and ethical guidance:
- Identifying a single authoritative regulator is difficult because ethical issues are far from clear.
- Some discrimination issues may be easier to remediate in AI (“AI will be easier to reprogram to be less discriminatory than people are”).
- Conclusion: AI requires strong policy guidance and experimentation to induce necessary changes.

Source: Embracing Artificial Intelligence by Joshua Gans, F&D Magazine, September 2023.

F&D: Cafe Economics: Embracing Artificial
- F&D: Cafe Economics: Embracing Artificial (Markdown version)
- F&D: Cafe Economics: Embracing Artificial (PDF)