AI Frontiers: Jesper Hvirring Henriksen (OpenAI DevDay)
TLDRBe My Eyes, in collaboration with OpenAI, introduces Be My AI on GPT-4V, enhancing visual assistance for blind and low-vision users. The feature allows users to receive AI-generated descriptions of images, improving accessibility to digital content and physical environments. With over half a million users and seven million volunteers, the app provides independence and choice, garnering overwhelmingly positive feedback. The technology's success is celebrated, highlighting its potential to revolutionize assistive technologies.
Takeaways
- 🌟 Be My Eyes, in partnership with OpenAI, has launched a new feature called Be My AI on GPT-4V to assist blind and low-vision individuals.
- 👥 The Be My Eyes community consists of over half a million blind and low-vision users supported by more than 7 million volunteers.
- 📞 The traditional model of assistance involves video calls where volunteers lend their eyes to those in need.
- 🚀 The introduction of Be My AI provides users with a 24/7 AI assistant alternative to human volunteers, enhancing independence and convenience.
- 🖼️ GPT-4 vision enables computers to 'see' and describe images, both in the physical and digital world, making media and photos accessible.
- 🌐 The application of GPT-4 vision transcends basic accessibility by providing detailed descriptions of images, including those in group chats and on websites.
- 🎯 The AI's ability to describe images is not only accurate but also witty and human-like, offering a more engaging user experience.
- 📈 Positive feedback from beta testers like Caroline and Lucy Edwards has been overwhelming, showcasing the life-changing impact of the technology.
- 📊 Since the launch of the beta in March, Be My AI has provided a million image descriptions per month with a 95% satisfaction rate.
- 🌐 Language support has been expanded to 36 languages by instructing the model to respond in the user's language, demonstrating its adaptability.
- 🤖 The integration of Be My AI into enterprise customer support, such as Microsoft's Disability Answer Desks, has significantly reduced the need for calls.
Q & A
What is the main purpose of Be My Eyes?
-The main purpose of Be My Eyes is to provide blind and low-vision people with access to visual assistance through a community of volunteers via an app.
How many blind and low-vision users does Be My Eyes currently support?
-Be My Eyes currently supports over half a million blind and low-vision users.
What was the motivation behind developing the Be My AI feature on GPT-4V?
-The motivation behind developing the Be My AI feature was to provide users with an alternative to human assistance, offering them an AI assistant available 24/7 that can see for them.
How does GPT-4V enhance the capabilities of computers in relation to vision?
-GPT-4V enables computers to see and provide descriptions of images, both in the real and digital world, making a vast range of applications accessible that were previously not possible.
What is a common issue with media on apps and websites for visually impaired users?
-A common issue is the lack of meaningful alt text for images, making them inaccessible to visually impaired users who rely on screen readers.
How does Be My AI assist with inaccessible images on websites?
-Be My AI provides thorough descriptions of any image encountered online or in an app, making content with missing or insufficient alt text accessible to visually impaired users.
What was the reaction of one of the beta testers, Caroline, to the Be My AI feature?
-Caroline, a beta tester, significantly increased her use of the service, going from making about two calls a year to completing more than 700 image descriptions.
How has the Be My AI feature impacted user satisfaction according to the feedback received?
-The feedback has been overwhelmingly positive, with satisfaction ratings over 95% when excluding downtime and system errors.
What is one example of how Be My AI has been integrated into enterprise customer support?
-Be My AI has been deployed into Microsoft's Disability Answer Desks, where users can start with a chatbot instead of a call, with 9 out of 10 users not escalating to a call.
What future improvements in accessibility do AI models that can see and hear hold?
-AI models that can see and hear, and understand human speech, are believed to profoundly improve accessibility and assistive technologies in the future.
Outlines
🌟 Introduction to Be My AI and Its Impact
This paragraph introduces Jesper from Be My Eyes, a platform that connects blind and low-vision individuals with sighted volunteers via video calls for visual assistance. A new feature, Be My AI on GPT-4V, has been launched in partnership with OpenAI. The platform's mission is to provide visual assistance to those with limited vision, and it has grown to support over half a million users with the help of more than 7 million volunteers. The introduction of the AI assistant aims to offer users an alternative to human assistance, ensuring they can maintain their independence and not feel like a burden. The GPT-4 vision capability allows the AI to understand and describe images, which can be incredibly useful in both the physical and digital worlds. The summary highlights the potential of this technology to transform the lives of users by providing detailed descriptions of images, including those in group chats, websites, and other digital media. The introduction of Be My AI has been met with overwhelmingly positive feedback, and the platform continues to innovate and improve accessibility for its users.
📸 Daily Life Applications of Be My AI
This paragraph showcases the practical applications of Be My AI in the daily lives of blind and low-vision individuals. It follows Lucy Edwards, a beta tester, as she demonstrates how the technology assists her in various tasks such as cooking, doing laundry, and identifying items during a meal. The detailed descriptions provided by Be My AI enable users to understand their surroundings and ensure they are performing tasks correctly, such as checking for eggshells in eggs or reading expiration dates on products. The feature also helps users navigate social media by describing the content of images, thus making it more inclusive. The paragraph emphasizes the positive impact Be My AI has had on users' lives, as it provides a level of independence and convenience that was previously challenging to achieve. The success of the beta testing and the enthusiastic user feedback underscore the transformative potential of this technology in enhancing accessibility and quality of life for the visually impaired community.
Mindmap
Keywords
💡Be My Eyes
💡GPT-4V
💡Visual Assistance
💡Independence
💡Accessibility
💡Alt Text
💡Digital Use Cases
💡Human-like Responses
💡Beta Testing
💡Language Support
💡Enterprise Customer Support
Highlights
Be My Eyes launched a new feature, Be My AI on GPT-4V, in partnership with OpenAI.
Be My Eyes aims to provide visual assistance to blind and low-vision individuals through a community of volunteers.
The app has over half a million blind and low-vision users supported by more than 7 million volunteers.
Volunteers assist users through video calls, but the new AI feature offers an alternative for users seeking independence.
Users expressed a desire for an AI assistant available 24/7, prompting the development of Be My AI.
GPT-4 vision enables computers to 'see' and interact with the physical world, opening up a million applications.
Be My AI can describe images, including those on websites and in apps, making content more accessible.
The AI can provide thorough descriptions of images, such as photos in group chats or inaccessible images on websites.
GPT-4 Vision is not only accurate but also displays wit and human-like responses.
Beta tester Caroline increased her usage from two calls a year to over 700 image descriptions with Be My AI.
Lucy Edwards, another beta tester, demonstrated how Be My AI assists in daily life in a video.
Be My AI helps users with tasks like checking eggs for shells, reading expiry dates, and identifying laundry colors.
The AI can analyze and describe the contents of a meal, enhancing the dining experience for users.
Users can have Be My AI describe the scent of a product, such as perfume, by analyzing an image of the packaging.
Be My AI can assist with social media, describing uncaptioned photos so users don't miss out.
Feedback from beta testers has been overwhelmingly positive, with high satisfaction rates reported.
The feature was rolled out to a small Beta group in March and later to all iOS users, resulting in a million image descriptions monthly.
Support for 36 languages was added by simply instructing the model to respond in the user's language.
Be My AI has been integrated into enterprise customer support products, such as Microsoft's Disability Answer Desks.
9 out of 10 users who start with a chatbot do not escalate to a call, showcasing the effectiveness of Be My AI.