Approximately nine in ten organizations report using AI after the emergence of Generative AI but training, policy and public confidence are moving very slowly in comparison. Here are five charts to spot a few gaps.

Chart 1 — In just a couple of years, AI adoption jumped from half of organisations to almost all of them

During the pre-ChatGPT era surveys showed steady AI adoption. Then ChatGPT launched in November 2022 and the following survey showed an upward trend in AI adoption. By 2025, almost nine in ten organisations worldwide are reported to be using AI.

A line chart showing the percent of organisations reporting AI adoption from 2021 to 2025, across different regions. All lines revolve around 50–60% until 2023. After that they rise sharply due to GenAI boom. By 2025, every region sits between 82% and 91%.

Chart 2 — The AI Duopoly

Who is building this technology? Almost all of it comes from a couple of countries. The US is leading the race in terms of investments and production of large scale models followed by China. Everyone else, including Australia, is a spectator in comparison.

A scatter plot of 40 countries. X-axis shows total private AI investment 2013–2025 (log scale). Y-axis shows the number of large-scale AI models produced (log scale). Bubble size shows 2025 investment. Color shows world region. The US and China sit far above and to the right of every other country. Australia sits around the middle with about $10 billion invested and one large-scale model.

Chart 3 — Clerical workers are still most affected by AI, but software and web professionals are quickly becoming more affected

The International Labour Organization scored every occupation in the world’s standard job classification in terms of the exposure to AI once in 2023 and in 2025. The 2025 index confirmed that clerical and typists, data-entry clerks, secretaries still remained most exposed. While the sharpest upward trend went to software, web and data occupations. As AI models gained coding and reasoning skills, these professionals moved into the front line.

A scatter plot of 427 occupations comparing AI exposure score in 2023 (x-axis) and 2025 (y-axis). Colors indicate exposure category from grey (not exposed) to deep red (highest gradient). Points above the dashed line saw their score jump in 2025. Points below it had a decline. Clerical occupations sit furthest right and in the deepest red exposure categories and slightly below the line. Software, web and data occupations sit well above the line which indicate a huge increase in exposure.

Chart 4 — In a span of a year, AI usage went from half of students to nearly all of them

UK based survey of students and AI (HEPI/Savanta), AI use for assessments jumped from 53% to 88% in a single year while only about a third of students had received any AI training.

A dumbbell chart of survey measures about students and AI. Each row is a measure. A grey dot marks the 2024 value and a red dot the 2025 value. “Used any AI tool” jumps from 66% to 92%. “Used generative AI for assessments” from 53% to 88%. “Received AI skills training” sits at only 36% in 2025.

Chart 5 — Concern about AI is steadily increasing, Australians are more worried than before

An Australian study asked people globally how worried they are about AI in 2022 and again after two years in 2024. Mean worry value (on a likert scale) rose in 16 of the 17 countries surveyed. Only China relaxed. The same pattern is observed in the responses of Australians as well.

A slope chart comparing mean worry about AI in 17 countries across two surveys in 2022 and 2024. Lines are colored by world region. Sixteen of seventeen lines slope upward. The 17-country average rises from 2.6 to 3.0 (Likert scale). Australia, the thick red line with diamond markers, rises from 2.8 to 3.0. China is the only downward line (2.6 to 2.4).

This story is not AI fearmongering. It is about understanding the gap between AI’s speed and our readiness to use it. The US and China built almost all the large models, risking that other countries will be locked out of the best AI tools. Office workers and tech workers face rising exposure, raising concerns in people. Students use AI for assessments far more than they are trained. So there’s still a wide gap on multiple fronts. We need new assessment formats, AI training and policy, support for transitioning from affected jobs, and sovereign AI capabilities. The goal is to capitalize on AI’s advances by closing the gap, not feel overwhelmed by it.

References

McKinsey & Company. (2025). The state of AI: Global survey. Retrieved via the Stanford AI Index and Our World in Data. https://ourworldindata.org/grapher/share-companies-using-artificial-intelligence

Quid. (2025). Private investment in artificial intelligence by country [Data set]. In Stanford HAI, Artificial Intelligence Index Report 2025. Retrieved via Our World in Data. https://ourworldindata.org/grapher/private-investment-in-artificial-intelligence-cset

Epoch AI. (2025). Parameter, compute and data trends in machine learning [Data set]. Retrieved via Our World in Data. https://ourworldindata.org/grapher/cumulative-number-of-large-scale-ai-systems-by-country

Gmyrek, P., Berg, J., Kamiński, K., Konopczyński, F., Ładna, A., Nafradi, B., Rosłaniec, K., & Troszyński, M. (2025). Generative AI and jobs: A refined global index of occupational exposure (ILO Working Paper No. 140). International Labour Organization. https://doi.org/10.54394/HETP0387

Freeman, J. (2025). Student generative AI survey 2025 (HEPI Policy Note 61). Higher Education Policy Institute. https://www.hepi.ac.uk/reports/student-generative-ai-survey-2025/

Gillespie, N., Lockey, S., Ward, T., Macdade, A., & Hassed, G. (2025). Trust, attitudes and use of artificial intelligence: A global study 2025. The University of Melbourne and KPMG. https://doi.org/10.26188/28822919