The human brain is the most powerful computer ever created and though most people have been trying to create a copy of the brain using Silicon and chips, a dedicated group of people has been actively working on creating computers/processors using living tissue. These computers are called Bio-Computers and recently there has been a major breakthrough in the field due to the work of scientists from the Indiana University Bloomington.
They have managed to grow lumps of nerve cells called Brain organoids from stem cells, each of these organoids contain about 100 million nerve cells. The team placed these organoids on a microelectrode array which sends electrical signals to the organoids and also detects when the nerve cells fire in response. They then sent 240 audio clips as sequence of signals in spatial patterns with the goal of identifying the speech of a particular person. The initial accuracy of the system was at about 30-40 percent but after being trained for two days (with no feedback being given to the cells) the accuracy rose to 70-80% which is a significant increase. The team’s paper on this project has been published in Nature Electronics
Brain-inspired computing hardware aims to emulate the structure and working principles of the brain and could be used to address current limitations in artificial intelligence technologies. However, brain-inspired silicon chips are still limited in their ability to fully mimic brain function as most examples are built on digital electronic principles. Here we report an artificial intelligence hardware approach that uses adaptive reservoir computation of biological neural networks in a brain organoid. In this approach—which is termed Brainoware—computation is performed by sending and receiving information from the brain organoid using a high-density multi-electrode array. By applying spatiotemporal electrical stimulation, nonlinear dynamics and fading memory properties are achieved, as well as unsupervised learning from training data by reshaping the organoid functional connectivity. We illustrate the practical potential of this technique by using it for speech recognition and nonlinear equation prediction in a reservoir computing framework.
This PoC doesn’t have the capability to convert the speech to text but this is early days and it is possible that with more fine-tuning we will be able to create a system that will allow us to to speech-to-text with a much lower power consumption than the traditional systems. However, there is a big issue of maintenance and long term viability of the organoids. Currently the organoids can only be maintained for one or two months before they have to be replaced which makes it difficult to imagine a commercial/home PC like deployment of these machines as the maintenance costs and efforts would make it unfeasible. On the other hand it is possible that we might see more powerful versions of this setup in research labs and data-centers which would have the capacity to maintain these systems.
I am looking forward to seeing more advances in this field.
– Suramya
Source: AI made from living human brain cells performs speech recognition