AI driven search.

[Mr Standfast]

a request to try an AI Driven search pops up when/ if you try to do an image search from the main page.

 

I've had a go and to be honest it's very early days and then it broke.

 

A search for Edward Colston produced a choice between a blank page and some random man, it then asked me which page I preferred.

 

Like I say early days.

[gvallee]

I wonder if CTR/Ranking will become redundant if AI search is used.

 

[John Mitchell]

I'm not seeing the pop-up box any longer. What exactly is an AI search? How does it differ from an ordinary one if the same search terms are used in both cases?

[gvallee]

2 hours ago, John Mitchell said:

I'm not seeing the pop-up box any longer. What exactly is an AI search? How does it differ from an ordinary one if the same search terms are used in both cases?

 

I'm not seeing it any more either. It only lasted a day or two. Curious to know how it works too.

 

 

[John Mitchell]

34 minutes ago, gvallee said:

 

I'm not seeing it any more either. It only lasted a day or two. Curious to know how it works too.

 

 

 

I asked ChatGPT:

 

How do AI stock photography searches work?

 

ChatGPT

AI stock photography searches typically work by leveraging machine learning algorithms and computer vision techniques. Here's a general overview of how the process works:

 

Image recognition and analysis: The AI system first analyzes the visual content of images using computer vision algorithms. It identifies various visual elements, such as objects, colors, shapes, textures, and even more complex patterns like facial expressions or scenes.

 

Metadata extraction: The system extracts relevant metadata from the images, which may include information like image dimensions, resolution, file type, and any available embedded tags or keywords.

 

Indexing and categorization: The extracted visual features and metadata are then indexed and categorized in a searchable database. This indexing process allows for efficient retrieval and organization of the vast collection of stock images.

 

Query processing: When a user initiates a search, the AI system analyzes the search query and attempts to understand the user's intent. Natural language processing techniques are often employed to interpret the query, extract keywords, and identify any specific requirements or constraints.

 

Similarity matching: The AI system compares the user's query against the indexed images using similarity matching algorithms. These algorithms evaluate the visual features and metadata of the query and try to find the closest matches in the database.

 

Ranking and relevance: The system ranks the retrieved images based on their relevance to the user's query. Various factors are considered, such as keyword relevance, image quality, popularity, recency, and user feedback. The aim is to present the most suitable and appealing images at the top of the search results.

 

Display and retrieval: The AI system displays the search results to the user, usually in the form of a grid or list of thumbnail images. Users can browse through the results, view detailed information about each image, and select the ones that best meet their needs.

 

Continuous learning and improvement: AI stock photography systems often employ techniques like reinforcement learning or collaborative filtering to improve the search results over time. User interactions, such as clicks, downloads, and feedback, are collected and used to train the AI model, enhancing its ability to understand and satisfy user preferences.

 

It's important to note that the specific implementation details of AI stock photography searches may vary among platforms and providers. However, the underlying principles of leveraging AI, computer vision, and machine learning to analyze, index, and retrieve images remain consistent.