The Touring Test

Unless you are living under a rock, you will know that generative AI is coming to your phone. Within the next few months we will see:

• new iPhone hardware with Apple Intelligence
• Google Pixel 9 with Gemini
• ChatGPT app with Advanced voice mode supporting more human-like conversations

AI is not just coming to existing phone devices, we will soon see generative AI operating experiences using other hardware form factors:

• Wearables including glasses – including augmented reality
• Humanoid robots – including robots that can walk and talk either within a single venue or more widely
• Autonomous vehicles – including vehicles that can operate immersive experiences

What does this mean?

• Audio tours are obsolete
• Phone based AI tour guides that are not as capable or as distributed as these device based AI services are also in deep trouble. If you are a vertical specialist (e.g. travel only) and you can’t beat the travel features provided by the general AI services such as consumers will find on their own phones you have no role to play in the future

BUT how much impact will these widely available digital tour guiding services have on human lead tour guiding? Will AI tour guides just be part of the consumer mix going forward, or will AI tour guides become the default mainstream experience including the leisure market and human tour guides become a luxury service at the low volume high price end of the sightseeing market?

How can we track technology progress to this inflection point?

Introducing The Touring Test

An AI tour guide can be considered equivalent to a human tour guide if customers achieve a comparable level of satisfaction, engagement, learning, personalisation, responsiveness, emotional connection, and storytelling before, during and after the tour.

• Satisfaction: The overall happiness and contentment customers feel about the tour experience
• Engagement: The level of interest and active participation of customers during the tour
• Learning: The amount of information and knowledge customers retain from the tour
• Personalisation: The extent to which the tour content, preferences, and route are tailored to the individual interests of customers
• Responsiveness: The speed and quality with which the guide answers customer questions and addresses their needs
• Emotional connection: The ability of the guide to create a sense of empathy and enthusiasm, fostering a personal bond with customers
• Storytelling: The effectiveness of the guide in narrating engaging, memorable stories that enhance the tour experience
• Error recovery: The ability of the guide to handle and recover from unexpected issues, such as running late, while maintaining a positive experience for customers

Guidelines

• Measurement flexibility: At this stage, we don’t need to define specific methods for measuring each characteristic. Entities claiming to have passed The Touring Test should use established methods for evaluating each characteristic and must publish their evaluation metrics to ensure transparency and consistency in their assessments
• Pass/Fail criteria: An AI tour guide will fail The Touring Test if it falls short compared to human tour guides in even a single characteristic
• Transparency: Unlike The Turing Test, it’s acceptable for customers to know that the tour guide is AI rather than human during the evaluation
• As the Touring Test name indicates, this specifically refers to tours, not activities or attractions. Additionally this test relates to operating tours, not designing itineraries or tour distribution & retail

As for my Autoura tech, we are getting there. I can see a product pathway to passing The Touring Test.

I dont think we need to pass in order to begin to scale up, but the industry scale flip to AI first tour guiding rather than human first tour guiding will not happen until then. Passing The Touring Test will be the inflection point for significant scale up (& disruption).

Generative AI is improving, our product is improving, non-phone devices are coming on the scene, maybe 12-24 months until we are talking about passing the Touring Test? (actual industry pivot point will lag)

Image: OpenAI