Copilot’s 71 minutes
GitHub randomly split 95 developers into two groups for a coding task. The ones using Copilot finished in 71 minutes; the ones without took 161.
The first year the money went down — and the first year adoption stopped climbing
Chapter 4 of the AI Index 2023 covers 2022, the year the decade-long investment curve finally bent. Private investment fell 26.7%, fewer companies were funded, fewer deals were done, and the share of organizations using AI slipped rather than rose. What did not fall was employer demand for AI skills, the depth of use inside companies that had already adopted, or China’s appetite for industrial robots. The numbers:
91.9global AI private investment in 2022 (US$ billions, down 26.7%)
189.6global corporate AI investment in 2022 (US$ billions, about a third below 2021)
47.4US private AI investment (US$ billions) — 3.5× China’s 13.4
50% of organizations using AI in at least one function (56% in 2021, 20% in 2017)
1,392newly funded AI companies worldwide in 2022 (1,669 in 2021)
268,200industrial robots installed by China in 2021 — 51.8% of the world total
4.2 — The decade’s first down year, and it was a steep one
Global corporate AI investment fell to $189.6 billion in 2022, roughly a third below 2021, and private investment dropped 26.7% to $91.9 billion. Both remain far above where the decade started: corporate investment has grown thirteenfold since 2013, and private investment is still 18 times its 2013 level.
Fewer deals, fewer companies
- There were 3,538 AI-related private investment events in 2022, a 12% drop from 2021 — but still a sixfold increase on 2013.
- The number of newly funded AI companies fell to 1,392 from 1,669 the year before. In 2013 the figure was 495.
- Every funding-size band shrank except the very top. Deals over $1 billion rose from 4 to 6, while the $500 million–$1 billion band collapsed from 13 to 5, $100–500 million fell from 277 to 164, and deals under $50 million fell from 2,851 to 2,585.
- The year’s largest private investment events were modest by the standards of the boom: $2.5 billion for GAC Aion New Energy Automobile, a Chinese electric-vehicle maker; a $1.5 billion Series E for Anduril Industries, a US defense products company; and $1.2 billion for Celonis, a business-data company based in Germany.
- The single biggest AI transaction of the year was an acquisition, not a funding round — Nuance Communications at $19.8 billion, followed by Citrix Systems ($17.18 billion) and Avast ($8.02 billion).
The United States still leads — and still fell 35.5%
The $47.4 billion of private AI investment in the United States was roughly 3.5 times the amount invested in China ($13.4 billion) and 11 times the amount invested in the United Kingdom ($4.4 billion). But leadership did not mean immunity: US private AI investment fell 35.5% year over year and China’s fell 41.3%. Aggregated across 2013–2022 the ranking is unchanged — the United States has attracted $248.9 billion, China $95.1 billion and the United Kingdom $18.2 billion, followed by Israel ($10.8 billion), Canada ($8.8 billion) and India ($7.7 billion).
- Entrepreneurial activity follows the same map. In 2022 the United States produced 542 newly funded AI companies, against China’s 160 and the United Kingdom’s 99 — 1.9 times the European Union and United Kingdom combined, and 3.4 times China.
- Over the decade the gap is wider still: 4,643 newly funded US companies, about 3.5 times China’s 1,337 and 7.4 times the United Kingdom’s 630.
- By focus area, the most funded categories in 2022 were medical and healthcare ($6.1 billion), data management, processing and cloud ($5.9 billion), fintech ($5.5 billion), cybersecurity and data protection ($5.4 billion) and retail ($4.2 billion).
- Most focus areas took in less money than in 2021. The ones that grew were semiconductors, industrial automation and network, cybersecurity and data protection, drones, marketing and digital ads, HR tech, AR/VR and legal tech.
- Perspective matters: even after the drop, most focus areas attracted more private investment in 2022 than they did in 2017.
4.1 — Investment fell; demand for AI skills did not
Across virtually every US sector for which there is data, the share of job postings asking for AI skills rose from an average of 1.7% in 2021 to 1.9% in 2022. The single exception was agriculture, forestry, fishing and hunting. Employers were still hiring for AI while investors were pulling back.
- Globally, the United States leads on the share of postings that require AI skills (2.05%), followed by Canada (1.45%), Spain (1.33%), Australia (1.23%) and Sweden (1.20%). Every country in the sample posted more AI jobs in 2022 than in 2014.
- By skill cluster, machine learning is the most requested (1.03% of all US postings), then artificial intelligence (0.61%), natural language processing (0.20%), neural networks (0.16%) and autonomous driving (0.15%). Every listed cluster is more in demand than it was ten years ago.
- Python is the specialized skill AI employers ask for most. It appeared in 296,662 US AI job postings in 2022, against 12,884 across 2010–2012, and its share of AI postings rose from 5.36% to 37.13% — a 592% increase.
- The fastest-growing skill shares belong to newer disciplines: Amazon Web Services (+4,763%) and data science (+3,767%), against +52% for Java.
The map of American AI jobs is spreading out
California posted 142,154 AI jobs in 2022, ahead of Texas (66,624) and New York (43,899). Measured against a state’s own postings the leaders look different: the District of Columbia tops the list at 2.95%, followed by Delaware (2.66%), Washington (2.48%), Virginia (2.42%) and California (2.21%). The interesting movement is in concentration. California still holds 17.87% of all US AI job postings, ahead of Texas (8.37%), New York (5.52%) and Washington (3.93%) — but its share has fallen steadily since 2019, and it no longer commands a quarter of the country’s AI jobs. All four of those states saw significant increases in absolute postings from 2021 to 2022.
Hiring, skills and the gender gap
- On LinkedIn’s relative AI hiring index, Hong Kong posted the strongest AI hiring growth in 2022 at 1.37, followed by Spain (1.19), Italy (1.18), the United Kingdom (1.18), the United Arab Emirates (1.15) and South Africa (1.13).
- Most countries in the sample now hire more AI talent than they did in 2016, but for many the hiring rate peaked around 2020, dropped, and has since stabilized.
- On relative AI skill penetration over 2015–2022, India leads at 3.23, then the United States (2.23), Germany (1.72), Israel (1.65), and Canada and the United Kingdom (both 1.54). A rate of 3.23 means Indian members are 3.23 times more likely than the global average to list AI skills across the same occupations.
- In every country in the sample, men’s AI skill penetration exceeds women’s. The highest rates for women are in India (1.99), the United States (1.28) and Israel (0.87) — against 3.27, 2.36 and 2.05 for men in the same countries.
4.3 — Adoption has plateaued; the companies already inside keep going deeper
50% of the 1,492 organizations McKinsey surveyed had adopted AI in at least one business unit or function in 2022 — down from 56% in 2021, but well above the 20% of 2017. Adoption has hovered between 50% and 60% for three years. What kept rising is how much AI the adopters use.
The average number of AI capabilities an organization has embedded in at least one function doubled from 1.9 in 2018 to 3.8 in 2022. The most widely embedded capabilities are robotic process automation (39%), computer vision (34%), natural language text understanding (33%) and virtual agents (33%). The most commonly adopted use case is service operations optimization (24%), followed by the creation of new AI-based products (20%), customer segmentation (19%), customer service analytics (19%) and AI-based enhancement of existing products (19%). Read together, those two lists describe a business AI that is broad but unglamorous — automation and text, pointed at operations and customers.
Where it pays, and where it is used
- On the cost side, the functions where most respondents saw decreases from AI adoption were supply chain management (52%), service operations (45%), strategy and corporate finance (43%) and risk (43%).
- On the revenue side, the leaders were marketing and sales (70%), product and/or service development (70%) and strategy and corporate finance (65%).
- By region, North America leads adoption at 59%, followed by Asia-Pacific (55%) and Europe (48%). The global average of 50% is 6 points lower than 2021, and Greater China fell 20 percentage points to 41%.
- The highest single industry-function combinations are risk in high tech and telecom (38%), service operations in consumer goods and retail (31%), and product and service development in financial services (31%).
- Movement was not all in one direction: the biggest year-over-year rise was strategy and corporate finance in consumer goods and retail (25 percentage points), and the biggest fall was product and service development in high tech and telecom (38 percentage points).
What business leaders say is holding them back
- The risks organizations consider most relevant when adopting AI are cybersecurity (59%), regulatory compliance (45%), personal and individual privacy (40%) and explainability (37%). The least cited are national security (13%) and political stability (9%).
- Fewer organizations act than worry. Only 51% take steps to mitigate cybersecurity risk, 36% regulatory compliance and 28% personal privacy — gaps of 8, 9 and 12 percentage points against the share that call those risks relevant.
- In Deloitte’s survey of 2,620 business leaders, 94% called AI important to their organization’s overall success and 82% agreed that AI improves performance and job satisfaction.
- 76% expected to increase AI investment in the next fiscal year — still a large majority, but 9 points below 2021 and 12 points below 2018.
- The top three challenges in starting AI projects are proving business value (37%), lack of executive commitment (34%) and choosing the right AI technologies (33%). The top barriers to scaling are managing AI-related risks (50%), obtaining more data to train a model (44%) and implementing the technology (42%).
- The outcomes leaders say they actually achieved: lower costs (37%), better collaboration across business functions (34%), valuable insights discovered (34%) and improved or customized products and programs (33%).
Five close-ups from the chapter
The details that do not fit into an aggregate — a controlled experiment on coding assistants, the year’s biggest transactions, what executives said out loud, and two shifts in what robots are for.
The year’s four biggest tickets
Nuance Communications at $19.8 billion was the largest single AI investment event of 2022 — and it was an acquisition, not a funding round.
What the Fortune 500 said out loud
268 Fortune 500 earnings calls mentioned AI in fiscal 2022 — down from 306 the year before, but still above 2018’s 225.
Robots designed to stand next to you
Collaborative robots — built to work with humans rather than for them — grew from 2.8% of new industrial installations in 2017 to 7.5% in 2021.
The robots that are not in factories
Professional service robot installations rose across hospitality, medicine, cleaning and logistics — with transportation and logistics up 1.5 times in a year.
Five questions about a year that turned
What the 2022 numbers do and do not say about AI’s place in the economy.
Did AI investment really collapse in 2022?
It fell hard, from a very high base. Global AI private investment came to $91.9 billion, a 26.7% decrease from 2021 and the first year-over-year drop in the last decade. Total corporate AI investment — mergers and acquisitions, minority stakes, private investment and public offerings combined — was $189.6 billion, roughly a third below 2021. But the ten-year picture is unchanged: private investment in AI was still 18 times its 2013 level, and corporate investment has grown thirteenfold over the decade. The retreat was broad rather than deep in any one place. There were 3,538 private investment events, down 12%, and 1,392 newly funded AI companies against 1,669 in 2021. Every funding-size band shrank except deals over $1 billion, which rose from 4 to 6.
Which countries are still putting money in?
The same ones, only less. The United States invested $47.36 billion in 2022, about 3.5 times China’s $13.41 billion and 11 times the United Kingdom’s $4.37 billion, then Israel and India (both $3.24 billion) and South Korea ($3.10 billion). Both leaders fell sharply — US private AI investment dropped 35.5% and China’s 41.3%. Cumulatively since 2013 the order is identical: the United States $248.9 billion, China $95.1 billion, the United Kingdom $18.2 billion, Israel $10.8 billion, Canada $8.8 billion. New company formation follows the same map. The United States produced 542 newly funded AI companies in 2022 against China’s 160 and the United Kingdom’s 99 — 1.9 times the European Union and the United Kingdom combined, and 3.4 times China.
If adoption has plateaued, is AI paying off for the companies that adopted it?
By their own reporting, yes. 50% of surveyed organizations had adopted AI in at least one function in 2022, down from 56% in 2021 but more than double 2017’s 20% — a plateau between 50% and 60% rather than a decline. Inside those organizations, use keeps deepening: the average number of AI capabilities embedded doubled from 1.9 in 2018 to 3.8 in 2022. On costs, respondents most often reported decreases in supply chain management (52%), service operations (45%), and strategy and corporate finance and risk (both 43%). On revenue, the leaders were marketing and sales (70%), product and service development (70%) and strategy and corporate finance (65%). The pattern in the chapter is a widening gap: adoption has stopped spreading, but the organizations already inside keep pulling ahead.
How dominant is China in industrial robots?
2021 was the first year in which China installed more industrial robots than the rest of the world combined — 268,200 against 249,000 for every other country. That was 5.7 times Japan’s 47,200 and 7.7 times the United States’ 35,000, with South Korea (31,100) and Germany (23,800) next. China overtook Japan in 2013, when it accounted for 20.8% of world installations; by 2021 that share was 51.8%. The global picture rebounded too: 517,000 industrial robots were installed worldwide in 2021, a 31.3% increase on 2020 and a 211.5% increase on 2011, while the operational stock rose 14.6% to 3,477,000 from 3,035,000. Virtually every country reported growth, led by Canada (66%), Italy (65%) and Mexico (61%); Singapore was the notable exception at -35%.
Where do the robots actually go, and what do they do?
Electronics first, cars second. Globally the sector installing the most industrial robots in 2021 was electrical and electronics (137,000), followed by automotive (119,000); every highlighted sector has installed more since 2019. By task, handling dominates — 230,000 robots were installed for handling in 2021, 2.4 times as many as for welding (96,000) and 3.7 times as many as for assembling (62,000), and every application category except dispensing and processing saw more installations in 2021 than in 2019. The two big economies differ in shape. In China the largest sectors are electrical and electronics (88,000), automotive (62,000) and metal and machinery (34,000), and every Chinese industrial sector installed more robots in 2021 than in 2019. In the United States automotive still leads at 9,800, but that sector fell year over year while food (3,400) and plastic and chemical products (3,500) rose.
The chapter in five lines
Headline findings from Chapter 4 · The Economy.
For the first time in the last decade, year-over-year private investment in AI decreased. Global AI private investment was $91.9 billion in 2022, a 26.7% decrease since 2021 — still 18 times greater than in 2013.
The demand for AI-related professional skills is increasing across virtually every American industrial sector — from 1.7% of job postings in 2021 to 1.9% in 2022.
In 2022, the $47.4 billion invested in the U.S. was roughly 3.5 times the amount invested in the next highest country, China ($13.4 billion).
The proportion of companies adopting AI has more than doubled since 2017, though it has plateaued in recent years between 50% and 60%. Organizations that have adopted AI report realizing meaningful cost decreases and revenue increases.
In 2021, China installed more industrial robots than the rest of the world combined.
Read Chapter 4 in full
Chapter 4 — 4.1 Jobs, 4.2 Investment, 4.3 Corporate Activity and 4.4 Robot Installations — with every figure, table and citation, is free from Stanford HAI.
Open the AI Index 2023 →