Seekr Games?
Seekr Games is a scavenger hunt platform that utilizes AI to enable instant task verification by taking pictures, or other various methods. With Seekr you can create your own scavenger hunts for your own events, whether that's a family reunion, or a team building activity, or a party you are throwing. Seekr Games has got you covered, and aims to make the experience of hosting a hunt as seamless as possible. Through our Scavenger Hunt editor, users have a couple of different options that can help them achieve just that, with a little bit of understanding. One is text descriptions, and the other is image checks.
What is a text description?
With Seekr your scavenger hunt is going to be powered by AI, that means we have to work with the AI a little bit to let it know what we want from it. To do this we need to try and describe what we want to the AI through the form of text, so put simply if we want the AI to validate that the picture a participant takes, and sends to be verified, is an apple. To accomplish this we can tell the AI we want an "apple".
Text description
with the text description we can provide AI with a general idea of what we are looking to find. If someone told you you had to find a apple. You would go and find an apple. This works with AI as well, though keep in mind, when you simply state apple, this leaves it vague and up for interpretation on what exactly an apple is. Sometimes you may to need to clarify to the Ai what kind of apple it's looking for. Otherwise just like how you might find a parade float of an apple, the Ai might accept a parade float apple as well.
Because this method can be descriptive, yet leave a lot of room for generalization or interpretation, we need to be careful how we describe something to the AI, if we want the actual fruit you can eat, adding verbiage such as a small real red delicious apple that comes from a tree. This is very descriptive, and can indeed help. But sometimes just like when we are given a task that has a lot of information, we tend to forget or get confused about things. AI can be the same way. We need to make sure that when we describe something, we are being short concise and straight forward.
Being general
Sometimes what we want is to be general. Scavenger hunts that provoke a Seekers creativeness and ingenuity can be quite engaging and fun. Sometimes we may want the user to find something like an apple. It doesn't have to be an apple, but if they can find something that still matches the description, like maybe they paint a painting, or create a sculpture of an apple. The AI can validate these as successful tasks, if we leave the validation up for interpretation.
Other ways being general are quite good
Allows hunts to be flexible for different locations
Enables hosts to move quickly without thinking of edge cases
Allows participants to get creative and problem solve a task
Allow for variation in responses, which can be quite fun, as hosts can see images sent by participants
As you can see providing general text descriptions can actually be a useful and fun, but sometimes we really are looking for a very specific answer, thats where text descriptions and or image matching, can really help you out.
Image match
Image matching is another form of verification the Seekr Games platform uses. This works by instead of describing what we want to the AI through text, we give it a picture, which as they say is worth a thousand words. In this case the AI does take it that way, images can be a great way to get the Ai to understand that you want a real apple, or a parade float apple. Without needing to specify the extra details. Providing an Ai with an image is just like if you were provided with an image, you could tell whether or not without much context whether two pictures meet a requirement.
Combined, granular control
The best and most granular use case, is combining both image match and text description. This allows users to create tasks that are very specific. Such as providing a picture of a sign, and stating in the text description, "sign". When you do a text description with image matching, think of describing the image by finishing the following text: does the image a participant provided contain the same [insert your text here] as the image the user/host provided. This is the context the AI gets when you describe and provide a picture. This enables the Ai to look for the same water bottle, or the same street sign. Something that gets down to a very specific location.
Conclusion
Ai verification is amazing, but remember that it has it's imperfections, it's not very good at face to face recognition, meaning it can't really tell people apart, that is something that humans are extremely wired and optimized to do. AI has not yet reached that level. But with these other methods we can get the control we need to create some really interesting and fun scavenger hunts! If you or someone you know is interested in creating a scavenger hunt, send them to SeekrGames to get started.