AI MUST READS — W22 2018, by City AI
Artificial Intelligence, Machine Learning and related fields are in a constant state of change. We want to inform but also encourage discussions on well presented topics we think are necessary in the context of putting AI into production. Every week we’re picking applied AI’s best articles plus adding a discussion starter
1. Microsoft is creating an oracle for catching biased AI algorithms
One of the constant companions of Artificial Intelligence at the moment is the plethora of questions about the ethical ramifications and about potential situations of bias creeping into the automation. If you’ve read almost any of my past must reads then you’ll know that this also isn’t new and there seems to be a new development of some kind in the “automation of bias” field almost weekly.
Microsoft and Facebook’s recent announcements SEEM to be the first practical steps being taken towards identifying and preventing this bias, that isn't to take away from the number of organisations set up and champions of the cause that have come forward because without these people I question whether these two leaders of industry would have taken these steps.
It does leave me questioning though. They can build an artificial intelligence system to detect A.I. in other artificial intelligence systems, if they can do that why can’t they prevent A.I. in their systems in the first place? Is this system for detection built by the same people that built the systems it will be searching for bias within?
Bonus video at the bottom of the page*
2. Why thousands of AI researchers are boycotting the new Nature journal
To have such an archaic profit-making model in place in a field that is at the cutting edge technology is an affront not only to the potential of the technology but also to the memory of politically charged internet activist Aaron Swartz and the many like him.
Machine learning is a young and technologically astute field… The community itself created, collated, and reviewed the research it carried out. We used the internet to create new journals that were freely available and made no charge to authors.
Information that has already been paid for once by the taxpayer and the work produced by the scientists and researchers should never be held and controlled by private institutions. This shouldn’t just be limited to AI and ML either, information of this type should always be free and available to the public and hopefully this will start a trend that will start towards the removal of restrictions.
3. Are you scared yet? Meet Norman, the psychopathic AI
This is idiotic and intentionally misleading with the simple aim to increase the foot traffic of this article. Do you know what sounds like a great idea? Lets demonize a technology that people are already worried about and perpetuate the stereotype that this technology will inevitably kill every single one of us.
In reality all they really created was a 16 year old child spouting off for the sheer shock value of what he has to say.
Not only does this entire article very quickly steer away from the whole story about the “Norman Bates A.I.” almost as if there was a word limit that had to be reached, but what it does say about the experiment that was performed is that in fact it has nothing to do with the specific AI that they’ve attempted to place human qualities onto, but instead it has to do with the data that it has been fed. I imagine that a child raised that had consumed the same imagery as Norman would have a similar out look. What a waste of time. This is how not to do it.
Bonus Article
Microsoft + GitHub = Empowering Developers
Although its not directly related to A.I. I feel like I’d be missing a trick if I didn’t talk about Microsoft’s announced acquisition of GitHub. Personally I’m not a big GitHub user (Don’t worry, I’m learning) and as such this announcement didn’t garner the visceral reaction from me that it did many people. Usually I’m the first to understand that the internet isn’t always the best place too get balanced, calm discussions on a topic and when I first saw the backlash I made the assumption it was the typical loud minority, but having read into it more there seems to be some genuine concerns.
There are a lot of other companies that could have acquired GitHub that would worry me twice as much as having Microsoft acquire the company, but still Microsoft will inevitably find themselves with conflicts of interest, how will they deal with these? Will they remain “all-in on open source” as they claim or will they exploit GitHub, driving away its user base?
Bonus Resource
UC Berkeley has open sourced the largest and most diverse self-driving dataset for the general public. It is being called ‘BDD100K’ and comes added with rich annotations.
Francesca Rossi, IBM Research & University of Padova, on unbiased AI. See also http://ibm.biz/five-in-five #AIethics
Join 6,000+ AI practitioners from over 100 countries at WorldSummit.AI this October!
