AI Video Ads: Why TikTok Viewers and Shoppers Still Resist

Cheap to Make, Hard to Sell

Generative AI has made video content dramatically cheaper to produce. The audience's willingness to watch it has not kept pace. That gap was laid out plainly by TikTok's chief executive, Shou Zi Chew, at the DBS Asia Future Forward conference in Singapore on October 12, 2026. He said AI-generated videos attract a smaller share of views on TikTok than their share of uploads would suggest 5. By his account, AI content currently draws only a low-single-digit percentage of the platform's views, even as experimentation with the tools grows 5.

"People still want to see very real, very authentic things," Chew said. He added that viewers are pushing back against videos that look obviously artificial 5. Coming from the head of a platform built on algorithmic video, that is a notable admission. It suggests that falling production costs alone will not buy attention.

A Broader Caution From the Same Stage

Chew's remarks on content were part of a wider conversation about AI's risks and timelines. He argued that cybersecurity dangers are "quite real." He warned that users often grant agentic AI systems permissions they do not fully understand, such as handing an agent a credit card and ending up with far more purchases than intended 1. He said the area needs "a lot of education" 1.

He also described a familiar pattern in technology adoption: the world tends to overestimate AI in the short term and underestimate it over the long run. Human adoption friction slows near-term progress, but that delay should not be read as a sign of limited long-term importance 2. His point about view counts fits this framing. Audiences have not yet warmed to watching AI content, which is a different claim from saying they never will 5.

What the Ad Research Shows

Two recent studies give Chew's anecdote more weight, though they do not fully agree with each other.

Research published in Harvard Business Review, co-led with Ipsos, showed 3,000 U.S. consumers 20 video ads for major brands 3. Half were made by people. The other half were produced entirely by AI, from concept through post-production, working from creative briefs the researchers derived from the original spots 3. Viewers had trouble telling the two groups apart. Even so, the AI versions scored worse on both short-term sales potential and long-term brand equity 3. The authors argue that the main danger is not obviously poor creative. It is work polished enough to get approved that then quietly underperforms, so they recommend treating AI creative as a governance and judgment question rather than simply a production tool 3.

A study from NYU and Emory University reached a partly different result. Ads generated entirely by AI significantly increased click-through rates. Ads where AI merely edited a human expert's design showed no such gains 4. Its most striking finding concerned disclosure: labeling an ad as AI-generated reduced clicks by 31 percent 4.

Reconciling the Signals

At first glance, one study says AI ads underperform and the other says AI-built ads win on clicks. The two are measuring different things, though. The NYU–Emory work focused on click-through, the immediate response to an image 4. The HBR–Ipsos study assessed video ads on sales potential and brand equity, which are deeper and longer-lasting outcomes 3. A piece of creative can grab a click without building much lasting value. Reading the two together, AI may be good at producing attention-grabbing surface material but weaker at the strategic craft that turns attention into durable brand preference.

The disclosure finding links the research back to TikTok. If an AI label alone cuts engagement by nearly a third 4, and viewers resist content that looks synthetic 5, then the authenticity penalty is real and measurable. That creates an uncomfortable tension. Platforms and regulators are moving toward labeling AI content, and marketers have a clear commercial reason to wish they weren't.

The Takeaway

The evidence points to a consistent conclusion. AI's main contribution to video so far has been lowering supply costs, and supply was never the scarce resource. Attention and trust are. On TikTok, AI content is plentiful but rarely watched 5. In advertising, it can pass for human work while delivering less 3, and it loses ground once people know where it came from 4.

For brands, cheaper generation does not remove the need for human strategic judgment. It makes that judgment more important, because acceptable-looking output is now easy to produce. The near-term winners are likely to use AI to speed up execution while keeping people in charge of the ideas, and they will plan for a world where audiences can see an AI label.