While we call our personal assistants, which are cloud-connected, a form of AI, in fact these AIs aren’t really “intelligent.” They are scripted systems tied to the web with decent speech to text interfaces. The responses they provide, with some exceptions (like Amazon’s experimental conversation feature), are identical to what you would typically get if you typed in your request using a browser on a search engine.
A true AI would be better able to not only determine your intent but to provide additional information that you also needed and automatically execute complex processes to give you what you need.
Currently, what we need is a reliable way to do in-home virus testing, to determine if what we are reading and hearing is fact or fiction, and to provide personally customized advice on how to stay both safe and employed during these hard times.
Intel just released demonstrated Pohoiki Springs, a highly powerful platform for both learning and inference designed to go into a cloud deployment at scale. While they demonstrated its ability to determine, with the proper sensors and software, since that same capability with other sensors could also likely decide whether or not you had a virus.
Let’s talk about that this week.
The Need For A Highly Differential AI
Determining scents is very difficult to do with technology. If it weren’t, we’d be deploying robots and sensors rather than dogs for explosives, illicit drugs, and locating buried people and bodies.
That same level of differentiation is being handled by automation in labs, but it hasn’t been scaled up and down to a point where it could be used in homes. There is an effort in Italy using an interesting partnership to create test kits that would work like pregnancy kits, cost a dollar, and be widely deployable, but the would only likely work with the current strain of Covid-19, not the others that are already cropping up.
What we need is something that can be updated electronically to determine if we are sick and not yet showing symptoms so we can medicate early and avoid spreading what we have to others, so that if we don’t need to go to the hospital, we can instead medicate at home. This approach would save user costs, reduce hospital overcrowding, and reduce the load on the insurance industry, potentially lowering rates.
Besides, companies, airlines, hospitals, government buildings, hotels, restaurants, and particularly cruises could use this during an outbreak to rapidly test people attempting to use their services and prevent those that are sick for spreading the illness to others.
Broad testing is seen as critical to containment and mitigation for any widespread sickness event and indeed essential for preventing the next Pandemic. It appears, thanks to faster travel, that pandemics are happening at a quicker cadence, so developing sensors and AIs that can rapidly identify them are not only critical to our health but to our economy, as it currently being globally demonstrated.
Pohoiki Springs’ Potential Impact
Pohoiki Springs is a highly capable differential AI that is potentially able to scale to Cloud workloads. Using an advanced Neuromorphic Design, it has the potential intelligence of a small mammal.
Now it may be more cost-effective if used in a Virus detection solution to primarily use this platform as a learning system. Then tying it to inferencing platforms for diagnosis, but that would limit the solution’s ability to provide highly customized results for users, which may be necessary during a virus outbreak.
For instance, while finding out if you have the virus is essential, the advice you get if you are in one of the exposed age groups already have another illness or are taking medication that impacts your immune system would be very different. Besides, your proximity to appropriate medical resources and their availability would also affect real-time advice.
Given, to make this work economically, the solution would also need to provide other capabilities than just identifying a virus. Say identifying a gas leak timely, listening to sounds of distress from a human or a pet, identifying and alerting on fake news, or an attempt to take advantage of you would all be potential uses for this AI.
Tied to on-body sensors (think Smartwatch) could improve the accuracy of apps that identify a stroke or heart attack and could even provide better early warning of medical events like an insulin imbalance, mental infirmity, or impairment due to substance abuse. The result would be a valuable system to individuals, medical facilities, and insurance companies that could provide a kind of protective shield around the user, reducing dramatically medical emergencies and improving the quality and timeliness of the response.
Wrapping Up: The Critical Nature Of True Cloud-Based AI
There are a lot of companies working on technology like this. NVIDIA was the first to understand the need for a broad market AI truly and initially focused on machine autonomy but has since migrated their solution to other problems. IBM was the first to bring an enterprise-class solution to market with Watson, and IBM’s is famous for reinventing itself over a century to remain relevant during changing times.
This latest technology from Intel adds to this base, and the more firms we have working on this technology, the safer we will be in the future. Appropriately applied in the Cloud, this capability could make Pandemics a thing of the past.