70+% of Players Don’t Recognize AI Presenters

In a quietly revealing trial, live-game operator BetGames tested whether AI hosts could stand in for human presenters on its online game-show streams.

The results matter for broadcasters, casinos, and anyone watching how generative AI translates into live entertainment.

Play your Favorite Slots Online with no Dress Code!

Find a complete casino overview here. Click on the casino of your choice to register.

The BetGames Experiment

The operator ran an A/B-style experiment on one of its live game shows in which human presenters were secretly replaced by real-time AI avatars.

The avatars recreated the presenters’ visual likeness and produced synthesized speech and lip-sync in the live stream.

Players weren’t informed, and normal gameplay continued while the company monitored both detection and player behavior.

How the AI presenters worked

Technically, the system combined several components to produce a convincing “digital human” in real time:

• A visual avatar modeled off a human presenter and animated to match facial movements.
• Text-to-speech and voice-synthesis for the host’s spoken lines.
• Real-time lip-sync and gesture animation tied to the generated audio.
• Integration with the live game engine so the avatar could manage pacing and interact with the game flow.

The trial revealed common live-digital-human challenges: occasional lip-sync mismatches, some unnatural phrasing from the voice synthesizer, and latency or translation hiccups when targeting multiple markets.

Key results of the experiment

The most striking metric: a large majority of players did not notice the swap. Over 70% of participants failed to detect that the host was synthetic.

Fewer than 30% recognized something was different. Measured KPIs — session length, average bet size, and number of bets placed — showed no statistically significant change when the AI presenters were on air. In short: players behaved the same.

Unexpected Conclusion: No Business Advantage (yet)

Despite the technical proof-of-concept, the company concluded there isn’t yet a clear commercial reason to replace humans.

Running synthetic hosts continuously still requires substantial compute resources and operational overhead.

At current costs and performance levels, the economics don’t yet tip in favor of fully automated presenters.

Why this Experiment Matters

Live-dealer and live-host products have long relied on human presenters to create a TV-studio atmosphere that keeps players engaged.

This test is one of the earliest public, real-world experiments showing that AI avatars can pass casual viewers at scale.

If the technology matures, operators could imagine 24/7 shows, highly localised hosts for specific languages, and flexible branding with fictional characters that don’t need payroll or time zones.

Broader Implications for the iGaming Industry

The trial points to both potential upsides and real risks:

• Possible advantages: endless scalability, instant localization, consistent branding, and the ability to prototype characters or formats quickly.
• Risks and limits: authenticity loss, regulatory and transparency issues (players may expect to know whether a human or AI is hosting), potential trust erosion, and the unresolved technical problems that show up in longer sessions.

Regulators and consumer-protection bodies will likely want clarity on disclosure and fairness if synthetic hosts become common.

A Key Insight: The Presenter May not Matter

Perhaps the most surprising takeaway came from leadership commentary: while AI worked and most players didn’t notice, the presence of an AI host didn’t increase engagement.

In other words, the presenter — whether human or synthetic — appears to be only one factor among many that determine player behavior in these formats.

The company suggested the better use of this technology may be to enable new types of games and experiences that humans can’t easily host, rather than simply replicating humans. The arising question then is: Do you care if you interact with real people on the internet?

Does anyone care?

This experiment is a clean, early example of how generative AI is moving from demos into live commercial environments.

The technical result — convincing digital presenters that escape immediate detection — is already real. The business result — whether that leads to replacing human hosts — is still unsettled.

For now, the technology looks like a powerful tool to expand creative options, not to make human presenters obsolete.

The company’s CEO Andreas Koeberl summarized the surprise succinctly: the AI “worked, and nobody cared” — a neat way of saying the experiment validated technical feasibility but didn’t (yet) prove a business case.

More Selected News For You

• Bet Bizarre (6) – A Flaming Flamingo fiasco
• Pragmatic Play passes €100 Million Drops & Wins Awards
• BGaming upcoming March Releases
• Metallica to Perform at The Sphere in Las Vegas
• KISS AI Live Casino – The AI Dealers of the Future
• Bet Bizarre – Naked Man In Casino Arrested For Murder
• Pragmatic Play introduces Free Bet Blackjack – An Analysis
• The Grand Finale: Top 5 Most Popular Branded Slots Revealed – Part 5