 | Source: (News Scientist) Quick Recap: Scientists at Indiana University Bloomington have achieved a rudimentary form of speech recognition using brain organoids, clusters of human brain cells linked to a computer. These organoids, resembling miniature brains, underwent training to recognize the voice of a specific individual from a collection of audio clips. The training involved 240 audio clips featuring eight people pronouncing Japanese vowel sounds. Initially, the accuracy was at 30-40%, but after two days of training, it improved to 70-80%. The organoids were placed on a microelectrode array, known as ‘Brainoware,’ which both transmitted electrical signals to the organoids and detected nerve cell activity. One interesting aspect of the study is the use of ‘adaptive learning.’ The organoids showed improved accuracy through repetition of the audio clips over two days, demonstrating the potential of unsupervised learning. The researchers believe that biocomputing systems, like Brainoware, could offer advantages over traditional AI, addressing issues such as high energy consumption and the inherent limitations of silicon chips. P.S. If you enjoyed this post you will love my newsletter where I talk about the latest AI developments. I know that it is hard to keep up in the AI world nowadays so I try to keep my readers up to date with the most interesting (and latest) information. [link] [comments] |
Predictive technology: computational social science
