AI cut the hours behind creative work and 70 percent of agencies did not change what they charge. That is not holding the line. It is a quiet discount. The client keeps paying the old rate for the part that got cheap, and pays nothing for the part that got harder, which is judgement. The billable asset moved. Almost nobody moved with it.
Start with the number, because it is the whole argument in one line. In the D&AD 2026 AI and Creativity Report, published 18 August 2026 and drawn from 197 creative leaders, 70 percent said AI had made parts of the creative process faster and they had not changed their pricing.
The industry read that as a win. Same fee, fewer hours, better margin. I have heard it framed as discipline, as refusing to let clients claw back value, as protecting the perceived worth of creative work.
It is not that. Look at what is actually inside the deliverable.
Two years ago, a campaign fee covered concepting plus a fairly large block of execution: the retouching, the variants, the cutdowns, the deck, the twelve aspect ratios nobody enjoys making. That execution block was the thing you could point at when a client asked what the money bought. It was visible, it took time, it clearly required people.
That block shrank. What did not shrink, and what nobody has been able to automate, is deciding which of the forty machine-generated options deserves to exist, and killing the other thirty-nine in a way you can defend to a client who paid for volume and now sees choice.
So the fee stayed flat, the cheap half got smaller, and the expensive half got bigger. The client is getting a discount on production they never asked for and a free ride on judgement they have never been charged for. Both sides think nothing changed. Everyone is wrong in a slightly different direction.
Why does everyone use AI and almost nobody think it is good for the industry?
Here is the tell that something is broken, and it is the most unbalanced pair of statistics I have seen this year.
In the Creative Boom State of the Creative Industry 2026 survey of 882 creative professionals worldwide, published 29 June 2026, 86 percent used AI tools in their work. In the same survey, 10 percent thought AI’s overall effect on the industry was positive. Fifty-eight percent called it mixed and 28 percent called it negative.
Eighty-six to ten. People are not ambivalent about a tool they use daily because they think it is bad at the job. They are ambivalent because the tool is doing the part that used to be countable, and the part that is left over does not show up anywhere: not in the timesheet, not in the scope, not in the fee, not in the credit.
The rest of that dataset behaves exactly the way you would expect if that were true. Fifty percent felt less financially secure than a year earlier, against 18 percent who felt more secure. Sixty-nine percent reported burnout in the past twelve months, rising to 77 percent among mid-career professionals, the exact group whose value used to sit in fast, reliable execution. Forty-eight percent said they were worried about where the industry is heading, with fewer than 38 percent confident. Thirty-eight percent were considering a job change.
Read that as a market signal rather than a mood. Half of a profession that just got dramatically more productive feels worse off. Productivity gains normally feel like something. These ones feel like nothing because they were priced at zero.
And the appetite for the shortcut has a ceiling that gets ignored in every keynote. The Uppbeat Creator Report 2026, fielded February to March 2026 across 1,792 creative professionals, found only 6.7 percent consider fully AI-generated content acceptable, while nearly 60 percent are willing to integrate AI tools into their workflows. The same report found that respondents aged 16 to 24 were the least likely group to use AI tools at all, which should trouble anyone who assumed the youngest cohort would simply automate the rest of us out of the room.
Sixty percent will use it. Under seven percent will ship it raw. The gap between those two numbers is a job description, and it is the job nobody has written into a contract.
I want to be careful here, because there is a lazy version of this argument that says creatives are simply underpaid and AI is the latest excuse. That is not it. Underpayment is a rate problem and it has existed forever. This is a description problem. The thing being sold in 2026 is not the thing named in the agreement, and both parties are operating off a document that quietly went out of date.
What exactly is the client paying for now?
Break the deliverable into its parts and the mismatch is obvious.
| Part of the work | What happened to it since 2024 | Who absorbs it now |
|---|---|---|
| Production and execution (variants, cutdowns, retouching, resizing) | Compressed. This is where the speed gain landed, per D&AD, 18 August 2026 | Client, as an unpriced discount inside a flat fee |
| Volume of options generated | Expanded sharply. Only 6.7 percent of creatives accept fully AI-generated output, per Uppbeat, February to March 2026 | Agency, unpaid, as review and rejection load |
| Deciding what should exist and what to kill | Became the scarce input | Agency, unnamed in scope |
| Defending the cut to the client | Became harder as visible output volume rose | Senior staff, off the clock |
| Fee | Unchanged at 70 percent of agencies, per D&AD, 18 August 2026 | Nobody, which is the problem |
Nothing in that table is about rates. It is about the fact that the line item everybody agreed on in 2024 describes a job that no longer exists in the same proportions.
The client-side numbers confirm this is not an agency-only phenomenon. The McKinsey Global Survey on the state of AI, fielded 4 May to 8 June 2026 across 1,719 participants in 97 nations, found nearly nine in ten organisations report regular AI use in at least one business function, 44 percent say AI is scaling across the enterprise (up from 38 percent a year earlier), and 56 percent use it in three or more functions, up from 51 percent. The most interesting figure in that survey for anyone selling creative services: 32 percent of organisations decided against buying at least one software product because they could build it internally with agentic coding tools.
If a third of your clients are now willing to build rather than buy the tooling, they will eventually ask the same question about the assets. The answer cannot be “our production is faster than yours.” That advantage has a half-life measured in months. The answer has to be the part they cannot generate, which is the call on what deserves to run.
Does Malaysia’s growing ad spend make this better or worse?
Worse, and this is where the argument stops being theoretical.
Malaysian digital advertising is expanding. The Digital ADEX Report from the Media Specialists Association with the Malaysian Advertisers Association and the Malaysian Digital Association, covering January to June 2025 and drawing on 21 agencies that represent about 60 percent of the country’s digital ad spend, recorded Q2 2025 digital adex of RM661 million, up 22 percent year on year. Q1 2025 came in at RM343 million, up 6.4 percent. Social media’s share of digital rose from 44 percent in Q2 2024 to nearly 50 percent in Q2 2025. Separately, more than 75 percent of all Malaysian advertising dollars now go to digital in 2026, according to Sandeep Mark Joseph, co-founder and CEO of Ampersand Advisory, in Marketing-Interactive. Forrester, in the same piece, expects marketers to cut display spend by 30 percent as AI and connected TV redefine engagement.
Put those together. Money is moving into the channel with the highest asset count per ringgit. Social eats half the budget, and social does not want one hero film. It wants fifty things, weekly, per market, per language.
So demand for volume is rising at precisely the moment volume became the cheap part. An agency that quietly absorbed the production discount in 2025 is now absorbing it across four times the asset count, with the review and rejection load scaling linearly while the fee does not. That is not a margin story. It is a slow structural bleed dressed up as growth.
Malaysia also gives us the clearest evidence that the real work relocated, and it comes from outside the industry. The Microsoft 2026 Work Trend Index, published 23 June 2026 and based on 2,000 Malaysian knowledge workers plus trillions of anonymised Microsoft 365 signals, found 92 percent of Malaysian AI users treat AI output as a starting point rather than a final answer. Sixty-nine percent say they produce work they could not have created a year ago. Twenty-four percent qualify as Frontier Professionals, against 16 percent globally.
Then the same study found that 19 percent say they are rewarded for reinventing how work gets done when it does not produce immediate results. And only 32 percent say leadership is clearly and consistently aligned on AI.
Ninety-two percent do the editorial work. Nineteen percent get rewarded for it. That is the agency problem stated in national statistics. The judgement layer is real, near-universal, and invisible to whoever signs off on value.
The behavioural detail underneath it is worth stealing. Among Malaysian Frontier Professionals, 57 percent pause to decide what should be done by a human versus an AI, compared with 39 percent of everyone else. Forty-two percent deliberately do some work without AI to keep their skills sharp, against 33 percent of others. Twenty-six percent report documented, repeatable agent workflows and quality standards, versus 18 percent of others.
Read those three lines again. The people getting the most out of the tools are the ones who spend the most time deciding when not to use them, and who write down what good looks like. That is not production skill. It is editorial standard-setting, and it is the thing that now determines output quality.
Is the problem that the work got worse, or that the valuable part went unnamed?
The second one, mostly.
There is a version of this conversation that blames the output. Everything looks the same, the machine flattens taste, the feed is full of smooth nothing. There is evidence for the slop half of it: Originality.AI’s July 2026 analysis of 5,000 public LinkedIn posts of at least 100 words classified 4,061 of them, or 81.2 percent, as likely AI, up from 53.7 percent in an earlier analysis of long-form posts by influential profiles across January to November 2025. Pangram’s scan of 1,002,627 posts between 24 April and 9 July 2026, in the same roundup, flagged over 40 percent of LinkedIn long-form posts as fully AI-generated.
But blaming the output lets the commercial structure off the hook. The work did not get worse because the tools are bad. It got worse in the places where nobody was paid to have an opinion about it. When production is the billable unit, volume gets attention. When production costs almost nothing, volume gets made anyway, and the only thing standing between a client and a hundred mediocre assets is someone whose judgement is not on the invoice.
That person exists in most agencies. They are usually senior, usually tired, and currently the least legible part of the P&L. The 69 percent burnout figure and the 77 percent among mid-career professionals in the Creative Boom data are what it looks like when a whole layer of work is performed without being described. Founders and studio leaders reported 59 percent burnout in the same survey, notably lower than the mid-career figure, which tells you the pressure is landing hardest on the layer that used to convert time into visible output.
There is a second cost that shows up later. When judgement is unpriced, it is also unstaffed. Nobody hires for it, nobody trains for it, and the juniors watch the senior work happen off the clock and conclude that it is not really part of the job. The Microsoft Malaysia finding that only 26 percent of Frontier Professionals have documented, repeatable quality standards, against 18 percent of everyone else, is a generous reading of an industry that mostly has none. Standards that live in one tired person’s head do not survive that person leaving.
So what actually changed about the job?
The unit of value moved from making the thing to deciding which thing should exist.
That sentence sounds soft until you notice that every dataset in this post points at it independently. D&AD found the process got faster while 70 percent of the commercial framing stayed still. Uppbeat found under 7 percent of creatives will ship raw machine output while nearly 60 percent will work with it, which means the entire profession has silently taken on an editing role. Microsoft found 92 percent of Malaysian AI users treat output as a draft. McKinsey found clients scaling AI into three or more functions and starting to build rather than buy. Creative Boom found the people doing all of this feel less financially secure than they did a year ago.
None of those are pricing findings. They describe a job whose centre of gravity shifted while its description did not.
The uncomfortable part is that the shortcut worked. Things really are faster. The reports that say quality slipped are also right, and both can be true, because speed was captured and judgement was not. You cannot compress the part of the work that was never counted in the first place.
An agency that keeps describing itself by what it produces is describing the cheap half. Malaysia’s ad market will happily buy that cheap half in enormous quantities, at rising volume, at a flat fee, until someone notices what is missing from the description.
The shortcut had a price. It just went on the wrong invoice.
Frequently asked questions
Did AI change what agencies charge in 2026?
Mostly no. The D&AD 2026 AI and Creativity Report, published 18 August 2026 from 197 creative leaders, found 70 percent of agencies had not changed their pricing despite AI making parts of the creative process faster.
Why do creatives use AI if they do not think it helps the industry?
Because the two questions are separate. Creative Boom’s June 2026 survey of 882 professionals found 86 percent use AI tools while only 10 percent call its effect on the industry positive, and 50 percent feel less financially secure than a year earlier.
What part of creative work got harder because of AI?
Deciding what should exist. Only 6.7 percent of creatives accept fully AI-generated content, per the Uppbeat Creator Report 2026, and 92 percent of Malaysian AI users treat AI output as a starting point, per Microsoft’s 2026 Work Trend Index.
Is demand for creative work falling in Malaysia?
No, volume demand is rising. Malaysian digital adex hit RM661 million in Q2 2025, up 22 percent year on year, with social approaching half of all digital spend, according to the Digital ADEX Report from the Media Specialists Association.
