close
Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2023 Mar 21;120(12):e2214840120.
doi: 10.1073/pnas.2214840120. Epub 2023 Mar 13.

Superhuman artificial intelligence can improve human decision-making by increasing novelty

Minkyu Shin  1 , Jin Kim  2   3 , Bas van Opheusden  4 , Thomas L Griffiths  4   5
Affiliations

Superhuman artificial intelligence can improve human decision-making by increasing novelty

Minkyu Shin et al. Proc Natl Acad Sci U S A. .

Abstract

How will superhuman artificial intelligence (AI) affect human decision-making? And what will be the mechanisms behind this effect? We address these questions in a domain where AI already exceeds human performance, analyzing more than 5.8 million move decisions made by professional Go players over the past 71 y (1950 to 2021). To address the first question, we use a superhuman AI program to estimate the quality of human decisions across time, generating 58 billion counterfactual game patterns and comparing the win rates of actual human decisions with those of counterfactual AI decisions. We find that humans began to make significantly better decisions following the advent of superhuman AI. We then examine human players' strategies across time and find that novel decisions (i.e., previously unobserved moves) occurred more frequently and became associated with higher decision quality after the advent of superhuman AI. Our findings suggest that the development of superhuman AI programs may have prompted human players to break away from traditional strategies and induced them to explore novel moves, which in turn may have improved their decision-making.

Keywords: artificial intelligence; cognitive performance; innovation; judgment and decision-making; novelty.

PubMed Disclaimer

Conflict of interest statement

The authors declare no competing interest.

Figures

Fig. 1.
Fig. 1.
Historical changes in quality and novelty of human decisions in Go. Panel A (Panel B) shows the fixed effect of each year (month) on decision quality along with its 95% CI, estimated using the median of Decision Quality Indices of all decisions made by each player in each year (month). Similarly, Panel C (Panel D) shows the fixed effect of each year (month) on novelty as measured with the Novelty Index, along with its 95% CI, estimated using the median of Novelty Indices of all games for each player in each year (month).
Fig. 2.
Fig. 2.
Yearly fixed effects on DQI, constructed from a dataset that excludes the move decisions that matched the counterfactual (optimal) decisions of the AI program. We find a similar increase in human decision quality following the advent of superhuman AI. The upward time trends of DQI after 2016 are still evident even when analyzing the restricted set of move decisions. This suggests that memorization hypothesis cannot fully explain the post-AI increase in decision quality.
Fig. 3.
Fig. 3.
Yearly fixed effects on Novelty Index, constructed from a dataset that excludes the move decisions that matched the counterfactual (optimal) decisions of the AI program. Even when this restricted set of move decisions were analyzed, we still find a similar increase in novelty following the advent of superhuman AI.
See this image and copyright information in PMC

References

    1. S. H. Choe, Google’s computer program beats Lee Se-dol in Go tournament. The New York Times, 15 May 2016. https://www.nytimes.com/2016/03/16/world/asia/korea-alphago-vs-lee-sedol.... Accessed 6 August 2022.
    1. McKinney S. M., et al. , International evaluation of an AI system for breast cancer screening. Nature 577, 89–94 (2020). - PubMed
    1. E. Cascetta, A. Carteni, L. Di Francesco, Do autonomous vehicles drive like humans? A Turing approach and an application to SAE automation Level 2 cars. Trans. Res. Part C: Emerging Technol. 134, 103499 (2022).
    1. Brown T., et al. , Language models are few-shot learners. Adv. Neural Inf. Process. Syst. 33, 1877–1901 (2020).
    1. Jumper J., et al. , Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021). - PMC - PubMed

LinkOut - more resources