AI Predictions for 2030: Expert Forecasts and a Track Record

Last updated: September 2026. Written by Josh Hutcheson, OnlineCourseing editor. Forecasts re-checked at the source (Gartner, World Economic Forum, IMF, IEA, FDA, Stanford AI Index) and our 2023 expert predictions graded on 22 September 2026. See our review methodology.

By Josh Hutcheson · E-Learning Specialist

Reviewing online learning platforms since 2019. Review methodology

THE SHORT ANSWER

Bottom line: the most credible AI predictions for the rest of the decade are about scale and change, not replacement: agents built into everyday software, more jobs reshaped than eliminated, fast-changing skills, and a surge in energy demand. Our own track record is a useful caution: of 25 predictions our expert panel made for 2023, 9 came true, 9 were partly right and 7 could not be measured.

Agents: in 33% of enterprise software by 2028, but over 40% of agentic projects canceled by end of 2027 (Gartner).
Jobs: 170 million created and 92 million displaced by 2030, a net gain of 78 million (World Economic Forum).
Exposure: almost 40% of jobs worldwide affected by AI, about 60% in advanced economies (IMF).
Energy: data center electricity set to more than double to about 945 TWh by 2030 (IEA).

Get ready with AI For Everyone →

How to read AI predictions

Before you spend money on the wrong online course, read this.

Get the free 2026 Platform Comparison Guide — 12 platforms compared on price, certificates, and refund policies. Instant PDF, plus my honest Tuesday picks.

No spam. Unsubscribe anytime.

AI forecasts have a mixed record. They tend to get the direction right and the timing wrong, and the loudest predictions are often the least specific. This page therefore does two things. First, it collects the forecasts for 2026 to 2030 that come from organizations with a published method and a named source: Gartner, the World Economic Forum, the IMF and the International Energy Agency. Second, it grades the 25 predictions our own expert panel made for 2023 against what actually happened, so you can see how well informed forecasters did.

For what is measurably happening today, rather than what might happen next, see our companion guide to AI trends in 2026.

What forecasters predict for 2026 to 2030

1. AI agents spread, and many projects fail along the way

Gartner predicts that 33% of enterprise software applications will include agentic AI by 2028, up from less than 1% in 2024, and that at least 15% of day-to-day work decisions will be made autonomously by agents by then, up from zero. It also predicts that more than 40% of agentic AI projects will be canceled by the end of 2027, because of rising costs, unclear business value or inadequate risk controls. Gartner warns of “agent washing”, where existing chatbots and automation tools are rebranded as agents, and estimates only about 130 of the thousands of agentic AI vendors are genuine (Gartner).

How to read it: both halves can be true. Agents are arriving in the software people already use, while many ambitious custom projects stall.

2. Jobs are reshaped more than erased

The World Economic Forum’s Future of Jobs Report 2025, based on a survey of more than 1,000 large employers covering over 14 million workers, projects that 170 million jobs will be created this decade, equal to 14% of today’s employment, while 92 million are displaced, a net increase of 78 million. The forecast covers all major trends together, including the green transition and demographic change, not AI alone. In percentage terms, AI and machine learning specialists, big data specialists and fintech engineers are among the fastest-growing roles (World Economic Forum).

The IMF offers a complementary view: AI will affect almost 40% of jobs worldwide and about 60% in advanced economies, where roughly half of exposed jobs may benefit from AI and the other half may see lower demand. Exposure is lower in emerging markets (40%) and low-income countries (26%) (IMF).

3. Skills keep changing fast

Employers expect 39% of key skills to change by 2030, down from the 44% they expected in 2023 but still a major shift (World Economic Forum). In the US, the Bureau of Labor Statistics projects employment of data scientists to grow 35% from 2025 to 2035, much faster than the average for all occupations (BLS). The implication is continuous learning rather than a one-time retraining.

4. Energy demand from AI more than doubles

The International Energy Agency expects data center electricity consumption to more than double from about 415 TWh in 2024 to around 945 TWh by 2030, slightly more than Japan uses today, with AI the most important driver. Investment in data centers had already nearly doubled since 2022, to half a trillion dollars in 2024 (IEA). Expect power supply, chip availability and cost to shape which AI products succeed.

5. Regulation phases in

The EU AI Act is taking effect in stages. Transparency rules have applied since 2 August 2026, requiring people to be told when they interact with an AI system and certain AI-generated content to be labeled (European Commission), and stricter obligations for high-risk systems follow later, on a timetable the Commission has proposed adjusting. Other countries are writing their own rules, and Stanford’s 2026 AI Index finds national AI strategies expanding, especially among developing economies (Stanford HAI).

6. The optimism gap persists

Stanford reports that 73% of AI experts expect AI to have a positive effect on how people do their jobs, compared with 23% of the public, a 50-point gap (Stanford HAI). Adoption at work may depend as much on closing that trust gap as on the technology itself.

The forecasts at a glance

Prediction	Source	Horizon	Our confidence
33% of enterprise software includes agentic AI	Gartner	2028	Medium: definitions of “agent” are loose
Over 40% of agentic AI projects canceled	Gartner	End of 2027	Medium-high: Gartner calls most current projects early-stage experiments
Net 78 million more jobs worldwide (all trends)	World Economic Forum	2030	Medium: employer survey, not a model of AI alone
Almost 40% of jobs affected by AI	IMF	Ongoing	High on exposure; impact per job varies
39% of key skills change	World Economic Forum	2030	High: direction consistent across surveys
Data center electricity about 945 TWh	IEA	2030	Medium: depends on efficiency and chip supply
Data scientist jobs +35% in the US	BLS	2025–2035	High: official projection

Future AI applications to watch

Beyond the headline forecasts, these are the areas where AI applications are already moving from pilots to real use, and where the next few years of change are most likely:

Science and drug discovery. Google DeepMind’s AlphaFold2 has predicted the structure of virtually all 200 million known proteins and has been used by more than two million people in 190 countries; its creators shared the 2024 Nobel Prize in Chemistry (Nobel Prize). Expect AI to shorten early-stage research further.
Medical imaging and diagnostics. The FDA’s list of AI-enabled medical devices authorized in the US has reached 1,614 entries, and 39 of the 50 most recent are in radiology (FDA).
AI agents at work. Assistants that plan and carry out multi-step tasks in office software, customer service and IT, within limits set by people.
Software development. AI coding tools that write, test and fix code, with developers moving toward review and design work.
Manufacturing and robotics. Predictive maintenance, quality inspection and robots; China leads the world in industrial robot installations (Stanford). See our guide to AI in manufacturing.
Education. Personalized tutoring and feedback; more than 80% of US high school and college students already use AI for schoolwork (Stanford). See AI in education.
Cybersecurity. AI on both sides of the fight, from deepfake fraud to automated defense. See AI in cybersecurity.

Scorecard: how our 2023 AI predictions held up

In late 2022 we asked 25 researchers, engineers and industry leaders what AI would bring in 2023. Their predictions are graded below against evidence available in September 2026. Roles are as given at the time. We grade on three levels: came true where public data clearly supports the prediction, partly where some of it happened or the timing or mechanism differed, and hard to measure where no reliable public data exists either way.

#	Prediction (contributor)	Verdict	Evidence in 2026
1	Conversational intelligence tools rise as meetings move online Sourabh Bajaj, tech lead, Google	Came true	AI assistants that summarize meetings and conversations are now routine; chatbots and assistants are the most widely scaled business use of AI, at 47% (McKinsey).
2	Machine learning runs in constrained devices and becomes more accessible Mitchell Spryn, research engineer, Facebook	Partly	Accessibility came true through open models (Stanford AI Index 2026), but the most capable AI runs in data centers, not on devices.
3	Governments open up data sharing for defense AI Robert Limmergard, Swedish Security and Defense Industry Association	Hard to measure	No public measure tracks defense data sharing.
4	Environmental and privacy scrutiny brings AI to lawmakers and courts Steve Meier, co-founder, KUNGFU.AI	Partly	Scrutiny of AI’s energy use arrived (IEA), but the landmark law came from the EU, with its AI Act, rather than from US courts.
5	No major breakthroughs in 2023, but much more R&D funding Prof. James Hendler, Institute for Data Exploration and Applications, Rensselaer Polytechnic Institute	Partly	The funding call was right (US private AI investment reached $285.9 billion in 2025, Stanford). The “no breakthroughs” call missed: 2023 was the year large language models went mainstream.
6	Open-source models spread; deepfakes keep plaguing the media Jack Hampson, CEO, DeeperInsights	Came true	Open-source AI contributions from the rest of the world now outpace Europe on GitHub (Stanford), and deepfake impersonation leads a 56% rise in AI-driven attacks (IBM, 2026).
7	AI, data and sustainability converge in finance Matthew Chan, ASIFMA	Partly	The sustainability debate did reach AI, mostly about AI’s own electricity use, now forecast to more than double by 2030 (IEA).
8	Embedded machine learning becomes the default for on-device analytics Adam Benzion, CEO, Edge Impulse	Hard to measure	On-device AI has grown, but no neutral figure measures whether it is the default.
9	Computer vision and edge deployments accelerate, especially in retail and healthcare AI team, Neal Analytics (now part of Fractal)	Partly	Vision drove healthcare AI: 39 of the 50 most recent FDA AI-enabled device listings are radiology. The edge share is not measured.
10	AI helps engineers predict network performance Energy Sciences Network (ESnet) researchers	Hard to measure	Widely used in network operations, but there is no public adoption measure.
11	More AI approved by the FDA, reaching beyond oncology Anant Madabhushi, PhD, Case Western Reserve University	Came true	The FDA’s AI-enabled device list now holds 1,614 entries, spanning cardiology, neurology and other specialties as well as imaging.
12	Symbolic AI combines with machine learning to improve human-machine interaction Jürgen Umbrich, Onlim	Partly	Human-machine interaction changed dramatically through chat, but large language models, not symbolic hybrids, drove it.
13	Defensive AI: infrastructure that actively defends itself Alex Bordei, MetalSoft	Came true	Organizations using security AI and automation extensively save USD 1.93 million per breach (IBM, 2026).
14	AI built into everyday software such as video calls, email and documents Reena Cruz, InvestinTech	Came true	Organizational AI adoption reached 88% (Stanford AI Index 2026), much of it through assistants inside familiar tools.
15	AI speeds up drug discovery screening Reece Armstrong, European Pharmaceutical Manufacturer	Partly	AlphaFold’s protein structure predictions won the 2024 Nobel Prize in Chemistry; AI-discovered drugs reaching patients is still early.
16	Financial institutions ramp up AI for ESG Jame DiBiasio, Digital Finance Group	Hard to measure	No public measure isolates AI in ESG work.
17	AI becomes augmented intelligence inside existing workflows Abhishek K, SublimeAI	Came true	80% of McKinsey respondents say AI improved their personal productivity, the assistive pattern this prediction described.
18	Transformers spread to every field; regulators demand explainability in critical systems Kemal Erdem, QuarkOwl	Came true	Transformer models now dominate, and the EU AI Act sets stricter rules for high-risk systems.
19	Machine learning used to find the business drivers companies can change Jason Glazier, PhD, Enterra Solutions	Hard to measure	Plausible and in use, but not measurable from public data.
20	Healthcare AI grows in the back office and imaging, not broad clinical care Charles Dinerstein, American Council on Science and Health	Came true	Imaging dominates approvals: 39 of the 50 most recent FDA AI-enabled device listings are radiology.
21	AI becomes part of everyday work and life, including the arts Brett Ashley Crawford, PhD, Carnegie Mellon University	Came true	Generative AI reached 53% population adoption within three years, faster than the PC or the internet (Stanford).
22	Biotech, precision farming and AI-integrated hardware lead innovation Kevin M. Purcell, PhD, Harrisburg University	Partly	Hardware became central (data center investment reached half a trillion dollars in 2024, IEA) and biotech had AlphaFold; farming is unmeasured.
23	AI, IoT and cybersecurity combine in smart highways and cities Anthony Davis, Highways.Today	Hard to measure	Deployments exist, but there is no reliable global measure.
24	Pandemic-era AI and voice applications for remote work Santiago Morante, PhD, Telefónica Tech	Hard to measure	Voice and meeting tools grew, but no clean measure separates this from wider AI adoption.
25	PC hardware and next-generation GPUs stay integral to AI Josh Covington, Velocity Micro	Partly	GPUs became the core of AI, but mostly in data centers: the US alone hosts 5,427 (Stanford).

Tally: 9 came true, 9 were partly right, 7 are hard to measure, and none were clearly wrong, although one part of one prediction, “no major breakthroughs in 2023”, plainly missed.

What the scorecard teaches

Direction beats timing. The panel correctly saw AI spreading through healthcare, security and everyday software. Almost nobody on it anticipated how quickly generative AI would reach the general public.
Specific predictions are more useful. The ones we could grade named something measurable, such as FDA approvals or deepfakes. Broad claims about “innovation” could not be checked.
Second-order effects surprised people. The sustainability conversation turned toward AI’s own power consumption, and the hardware story moved from PCs to vast data centers.
Use forecasts as scenarios, not schedules. Plan for several outcomes and build skills that stay useful across them.

How to prepare for the next five years of AI

Learn the basics now. Understand what current AI tools can and cannot do, and practise using and checking them in your own work.
Map AI onto your field. Identify which of your tasks are exposed, which could be augmented, and which rely on judgment and relationships.
Budget time for continuous learning. With 39% of core skills expected to change by 2030, short, regular learning beats occasional big retraining.
Follow measured sources. Stanford’s annual AI Index, the IEA and official statistics are better guides than vendor forecasts.
If you are technical, learn to build agents. Designing, evaluating and securing multi-step AI systems is where demand is heading.

Courses to future-proof your AI skills

DeepLearning.AI – AI For Everyone (Coursera). Andrew Ng’s seven-hour, non-technical course on what AI can do, how AI projects work and AI’s effect on society and jobs.
Google – AI Essentials (Coursera). Five short courses on using AI tools at work, prompting, responsible use and a final course on staying ahead of AI developments.
Udacity – Agentic AI Engineer with LangChain and LangGraph. For Python developers: build single agents, retrieval-augmented agents with a human in the loop, and multi-agent systems across three reviewed projects.

Both Coursera programs are included in Coursera Plus. Coursera removed its free audit option for most courses in 2025, so check the price or trial terms before enrolling. For more options, see our ranking of the best AI courses. For the specific fields these predictions touch, see our guides to computer vision, image processing, robotics, reinforcement learning, chatbot, IoT, embedded systems, predictive analytics and algorithmic trading courses.

See Google AI Essentials →

Frequently asked questions

What are the main AI predictions for 2030?

Credible forecasts point to AI agents built into a third of enterprise software by 2028 (Gartner), a net gain of 78 million jobs worldwide by 2030 as roles are both created and displaced (World Economic Forum, across all major trends), 39% of core skills changing by 2030, and data center electricity use more than doubling to about 945 TWh (IEA).

Will AI take my job?

More likely it will change it. The IMF estimates AI will affect almost 40% of jobs worldwide and about 60% in advanced economies, roughly half of which may benefit from AI rather than be replaced. The World Economic Forum expects more jobs to be created than displaced this decade across all the trends it studied, but with major shifts in the skills required.

How accurate are AI predictions?

Mixed. When we graded 25 predictions our expert panel made for 2023, nine came true, nine were partly right and seven could not be measured. Forecasters were good at direction, such as more AI in healthcare and security, and poor at timing, notably underestimating how fast generative AI would spread.

What will AI agents be able to do in the future?

Gartner predicts that by 2028 at least 15% of day-to-day work decisions will be made autonomously by agentic AI, up from 0% in 2024. It also warns that more than 40% of agentic AI projects will be canceled by the end of 2027 because of cost, unclear value or weak risk controls.

What are examples of future AI applications?

Leading examples include drug and protein research building on AlphaFold, medical imaging tools cleared by the FDA, AI agents that handle multi-step office and software tasks, AI-assisted coding, robotics in manufacturing, personalized tutoring, and AI-driven cybersecurity.

How can I prepare for the future of AI?

Build practical AI fluency now, including prompting and checking AI output, learn how AI applies to your own field, and keep skills current, since employers expect nearly two-fifths of core skills to change by 2030. Short courses such as AI For Everyone or Google AI Essentials are a low-cost starting point.

The verdict

The best-evidenced AI predictions describe a decade of rapid change rather than sudden replacement: agents in everyday software, a large reshuffling of jobs and skills, and heavy demand for computing power and electricity. Our scorecard shows why to hold forecasts loosely: informed experts reliably called the direction and regularly missed the timing. The practical response does not depend on which forecast wins. Build AI skills now, keep them current, and judge new claims by the evidence behind them.

See Udacity Agentic AI Engineer →