The drive for artificial intelligence needs to consider the purpose for having intelligence. Being clever for clever’s sake has limited practical applications.
This week, Mark Zuckerberg has been explaining the mission for Facebook going forward, focusing on three developments:
- Artificial Intelligence – because intelligent services will be more useful to consumers
- Virtual reality – because it is the next major computing and communication platform
- Internet.org – because it is the most basic tool people need to get the benefits of the Internet
We’ll save debating the reasoning behind each objective for another day (or maybe not). But comments about the first objective raised my hackles a little. According to the Facebook Q&A (via Engadget):
…Facebook’s in the midst of building AI systems “that are better than humans at our primary senses.” They’re designing one to be able to detect everything in an image or video: people, objects, animals, backgrounds and locations, among others. …it could, for instance, tell a blind person what [the image] is about.
First up, that example does not demonstrate an AI better than humans at their primary senses. At best, that involves catching up.
Second… really?! Better than humans at our primary senses? It’s another one of those arrogant statements that ignores how little we really understand about how the senses work or why we even have them. As if translating an image into sterile words matters more than the human experience behind it.
Because sterile words are all that computers will provide unless or until they achieve awareness. Not rules-based chess-playing logic but conscious awareness about what matters and what doesn’t.
I am technically partially sighted. However abandon me on a savannah without my contact lenses and I don’t need a high-res picture to make a decision if there is a shape vaguely like a lion in the distance. I’ll run/climb a tree first and save the scrutiny for later. A basic awareness of whether it is a big or small cat trumps the ability to identify the species.
And it turns out that computers are as easily fooled as us by optical illusions. Recent research by Cornell University found that computers detected objects with high levels of certainty within images that humans considered to be patterns or white noise. See the link under References and decide for yourself if you can spot the penguin…
But what really matters is why we have vision and what we use it for. What do you see when you look at the featured image at the top of this post? For me, it is more than what I see, it is what I remember. Walking along that beach, paddling in the water, enjoying the trip of a lifetime. What would be more memorable to a blind person. A sterile description of the objects detected in the image, or a person verbally recalling a memory and telling a story?
I love technology and how it can be used to advance and improve life. I despair how some technologists place so little value on the experience of life itself.
- Zuckerberg reveals Facebook’s AI, VR and Internet.org plans – Engadget
- Images that fool computer vision raise security concerns – Cornell University
- Can a robot become conscious?
- Self-learning will not create a brain
Featured image: New Zealand South Island, East coast (Author’s own photo)
So very well said! Excellent post.