How does the AI alt text tool work?

Training the model

AI alt text in the CMS is generated with a custom version of a major multi-modal large language model (LLM). Our custom prompts include: guidelines on accessibility and cultural sensitivity, as well as best practices for the world of arts and culture. The custom model is also trained on professional examples of alt text, written by experts at a top accessibility firm.

On top of that, we instruct the model to avoid common errors and edit its own output. Our goal was to create a model that could generate alt text that would satisfy the standards of not just accessibility experts, but you, our partners.

Testing and refining the AI tool

We built an experiment that would allow experts and amateurs alike to blindly rate a group of alt text for a wide range of art. This group included hundreds of examples of alt text written by non-experts, by the AI model, and “golden examples” written by accessibility experts. 

Each of the alt text examples was rated on a scale of 1 (unacceptable) to 5 (excellent) for each of these categories:

• Clarity
• Conciseness
• Accuracy
• Objectivity
• Relevance

In stage one, we internally evaluated the alt text samples, and identified systematic failures. The model over-interpreted images, lacked key spatial information, and inconsistently read pieces of text within images. We added targeted prompts to address those failures before stage two.

Stage two included the blind test and subsequent statistical analysis. Each image’s alt text was evaluated by three separate accessibility experts. Without knowing the source of the alt text, they rated the AI model’s alt text, text written by non-experts, and text written by experts. The results indicated that AI alt text samples were rated as equal to or better than the samples that were written by people, with 80% of the AI alt text rated as “Acceptable,” “Good,” or “Excellent.”

Stage three included a limited release to nine Connects partners. At the end of this limited release, all nine partners said they would continue using the AI. Most partners reported that the tool helped reduce the time and effort it took to write alt text for their images.

Using AI responsibly

The model is trained to avoid making mistakes. You may notice alt text that surprisingly omits a subject’s gender or race. The misinterpretation of gender and race can be harmful, and the two can’t always be visually determined. With that in mind, we trained the model to avoid mentioning a subject’s race, and only mention gender when it’s 100% sure. If you find that the alt text doesn’t fully represent key information, you can always add those details where they’re necessary.

Bloomberg Connects is committed to using AI tools responsibly and transparently. Your content remains confidential and secure within Bloomberg's cloud environment at all times. We do not allow third parties to train on, store, or publish your content. Our AI alt-text tool is fully opt-in, requires explicit partner approval before any content goes live, and is designed to give you complete control over how AI is used within your experience.