Page 463 - AI for Good Innovate for Impact
P. 463

AI for Good Innovate for Impact



                   Use Case 15: AI-Powered Universal Communication Platform:

               Breaking Down Barriers Through Collaborative Innovation                                              4.4-Productivity




















               Country: United Kingdom

               Organization:  The Scott-Morgan Foundation

               Contact Person(s): LaVonne Roberts, lavonne@ s cottmorgan foundation .org, +1 512 695-4500


               1      Use Case Summary Table

                Item                 Details

                Category             Productivity
                                     Over 500 million people worldwide face barriers to communication.
                                     Research shows only 18.2% of students needing communication
                                     support receive any form of Augmentative and Alternative Commu-
                Problem Addressed
                                     nication(AAC), with just 4.8% accessing speech-generating devices.
                                     Traditional assistive technologies exist in isolated silos-prohibitively
                                     expensive, medically classified, and lacking integration.

                                     Eye-tracking and adaptive input systems with Large Language Models
                                     (LLMs) for prediction Personalized voice synthesis (ElevenLabs) and
                Key Aspects of Solu-  avatar representation (D-ID) Open ecosystem approach connecting
                tion
                                     previously siloed technologies Consumer hardware platform reducing
                                     costs by 80%

                                     Ecosystem integration,Voice synthesis,Avatar generation, Eye- tracking,
                Technology Keywords
                                     LLMs Assistive technology
                                     Private - Personal voice samples and user interaction data (with strict
                Data Availability
                                     privacy controls)
                Metadata (Type of  Audio (voice samples), Image (eye-tracking data,facial expressions), Text
                Data)                (user communication)

                                     Voice models from limited samples LLMs for prediction Computer vision
                Model Training
                                     for eye-tracking












                                                                                                    427
   458   459   460   461   462   463   464   465   466   467   468