The AI revolution is gaining refinement..

[Raz0rEdge]

With recent advances of AI, we're seeing it being integrated into many many things. Siri, Alexa, "Hey Google" started us on a journey of doing things with our voices, and what started out as simple aggregators have gotten smarter.

Humane is trying to make AI accessible in a very simple way through their AI Pin.

I'm definitely excited to see how these technologies evolve over the next few years. As a long time fan of sci-fi, things that were literally fantasy about 20-30 years ago, might start slowly becoming reality.

I'm still not holding my breath for the Star Trek transporter to become a reality in my lifetime, but who knows.

Humane Ai Pin: Order Now

Screen-free, Hands-free, and App-free. The world’s first Ai wearable that lets you bring Ai with you everywhere and interact with the world around you.

hu.ma.ne

 

[IWT]

AI sure is gaining momentum. The only area in which I might claim some knowledge is medicine; particularly in relation to interpretation of images, most often radiological.

There are real difficulties to be recognised and dealt with. For example, if AI is used to read an image and comes up with a defined diagnosis - what if it's wrong and the examining doctor sides with the AI and the patient suffers; who carries the can? Not AI, it's a computer. Doctor to blame.

Image analysis; interpretation of an ECG/EKG; classification of tissues in a pathology exam etc.

AI is hear to stay; but the legal responsibility resides with the physician. The way forward is for AI to be treated as an adjunct to assist the doctor, but the ultimate responsibility lies with a human being. Take my word for this, it's a huge challenge for medics who take the blame if they are wrong, but get no credit when they get something right if the AI said it first.

Ian

 

[Raz0rEdge]

Ian,

Agreed. AI has a long way to go to make life threatening decisions. The way to think about this is (if you remember) the early 90's and the speech recognition. When Dragon Naturally Speaking came out, it was really bad at recognizing anything. The user had to spend countless hours talking to it so that it could recognize your speech as well as the words you were saying.

Compare that to the natural language systems now and it's leaps and bounds beyond in what it can recognize with little to no training and so on.

I worked on a veterinary diagnostic device that uses different types of "snaps" to test for different diseases. You put a few drops of some liquid and then insert it into this device, the device would read a small QR code to know what test was inserted, trigger the test and then watch the window where dots appear indicating the status of the disease being tested.

The first iteration that I built for this company was to do all that but leave the decision to the Vet based on a timer. The second iteration was to use computer vision and retain the results of the test on the screen (with a guess as to pos/neg result) for the Vet to validate. Eventually, with enough successes (like 90%+) the goal was to allow the machine to detect the results and record it as such with no Vet intervention/validation at all.

As far as blame goes, that's something we need to figure out right now and the human overseeing the AI is always the responsible party. Think of all the self-driving cars right now, if it causes/gets into an accident, the driver behind the wheel is responsible for not having prevented it. Negligence on their part is further penalized.