Inside the Briefcase

2017 State of Technology Training

2017 State of Technology Training

Pluralsight recently completed an in-depth survey of 300 enterprises...

IT Briefcase Exclusive Interview: Keeping Your (Manufacturing) Head in the Clouds

IT Briefcase Exclusive Interview: Keeping Your (Manufacturing) Head in the Clouds

with Srivats Ramaswami, 42Q
In this interview, Srivats Ramaswami,...

IT Briefcase Exclusive Interview: New Solutions Keeping Enterprise Business Ahead of the Game

IT Briefcase Exclusive Interview: New Solutions Keeping Enterprise Business Ahead of the Game

with Sander Barens, Expereo
In this interview, Sander Barens...

IT Briefcase Exclusive Interview: The Tipping Point – When Things Changed for Cloud Computing

IT Briefcase Exclusive Interview: The Tipping Point – When Things Changed for Cloud Computing

with Shawn Moore, Solodev
In this interview, Shawn Moore,...

Driving Better Outcomes through Workforce Analytics Webcast

Driving Better Outcomes through Workforce Analytics Webcast

Find out what’s really going on in your business...

How Open Source AI Effects the World Around Us

March 8, 2017 No Comments

Featured article by Michael Green, Chief Analytics Officer, Blackwood Seven

The marriage of artificial intelligence and the open-source concept is still very new, and we’re beginning to see its production come to fruition. After a few years, outsiders have been able to help play with other’s open-sourced AI technology and in turn, send these improvements back to the original creators.

This past December, Elon Musk’s OpenAI and Google DeepMind released sources codes, tool kits and algorithms for software developers and users to modify, and theoretically improve it. Through these methods of gamification in AI, open-sourcing the software allows researchers and developers to create their own levels and test their own AI on the platforms.

The OpenAI and DeepMind games create real-world solutions for AI to compete against bots, allowing the AI to learn and cooperate to solve problems.

By bringing the open-source concept to artificial intelligence, AI code has been opened up to the outside world. There are significant benefits to opening this power to so many – but there is room to be cautious in its development.

The Benefits of Open Source AI

Open sourcing this technology is important for several reasons. The primary benefit is that it provides researchers and engineers worldwide with a benchmarking tool and platform for evaluating intelligent agents and predictive models in a gaming universe.

Naturally it also extends the ability for the community to contribute with new features and/or bugs fixes.

Open Source and the Advancement of AI

Within the gaming domain for AI, having an open source framework will definitely speed things up as more people will be able to test their algorithms on the same conditions as everyone else.

Moreover, since the OpenAI Universe is based on games where the AI has the same capabilities and limitations as a human player, performance can be measured between all agents, both human and artificially intelligent ones.

How AI Impacts Humanity

There is no right or wrong answer to gauge AI’s impact on humanity. And therefore, defining its role becomes a difficult question to answer where there is no right or wrong reasoning. But I’ll give my 20 cents.

In order to get some context, we need to differentiate between different levels of AI. Currently it can be divided into:

1. Narrow AI

2. General AI

3. Super AI

Narrow AI is already a reality today. This is not a particularly scary scenario since it is already all around us. We all use it on a daily basis, and this is especially true for the cell phones we carry around. They know more about us than we usually realize. Even so, Narrow AI can only solve domain specific tasks such as optimizing click ratios for online ad banner buying, recognizing spoken language, categorizing certain types of images, knowing where you will typically go on a Sunday evening (Siri, Google Now et. al.), or even beat the world champion at Go.

Nonetheless, all of these Narrow AI would perform horribly if thrown into an environment that looked differently than the one they were operating in. In other words, they don’t generalize across domains.

Now, General AI, which the open sourcing of these platforms might be a small step towards, is a bit scary. These kind of AI would indeed generalize across domains and learn to do all things as well or even better than us humans.

General AI will be able to take the skills they learned during playing Go and apply them to another part of life, like business decisions and other related topics. Humans do this all the time

For example, we can take lessons from our latest Karate class and apply it in a non-violent way in a meeting room. As of today, this is not something an AI can do.

The later stage, which to most people is the really scary one, is where the Super AI comes into play. This AI is able to generalize its learnings and apply them in rocket-like speed. What’s even more impressive is that this type of intelligence is limitless. It doesn’t share the limitation of our biological brain and it doesn’t share our scaling problems.

Once Super AI takes off it will surpass humans in every possible way. It will advance science at a pace we’ve never seen before. It will realize problems and act upon them much sooner than we would have. It will be able to perform tasks we never believed were possible in the past.

For all purposes, to us humans, this AI will be a God. All that remains to be seen is: will it be a nice God?

About the Author

Michael Green, Chief Analytics Officer, PhD

Spawned in the Swedish Special Forces, bred in Theoretical Physics, hardened in Finance and softened in Marketing; Dr. Michael Green has developed, tested and deployed statistical machine learning models for a wide range of products over the last 15 years to help people make smarter and faster decisions with confidence.

He’s the inventor of the Artificial Intelligence engine deep within the Blackwood Seven AI media platform responsible for the quantification and decision making in their internal processes.

Dr. Green’s background is somewhat nontraditional in the media world as he has a doctorate in Theoretical Physics where his research was primarily focused on emerging complex systems. It didn’t take long before he realized the commercial potential for AI. He started out building Deep Artificial Neural Networks for helping physicians detect acute coronary syndromes in the emergency departments in Swedish hospitals saving 4% more lives while reducing costs by 40%. After that he spent a couple of years in the financial sector developing the statistical foundation for a business intelligence software that allowed banks and asset management companies to calculate the amount of money they stood to lose on a given trading day. He later traveled the EMEA region installing and evangelizing this software in major asset management companies. The last 7 years he’s been in the media and marketing industry developing and refining machine learning and AI methods together with Bayesian inference to quantify the effect different activities have on hard core KPI’s like sales.

In his spare time he enjoys running and has participated in multiple marathons over the years. In addition he teaches and practices Martial Arts which helps maintain a healthy work-life balance. Currently, he holds the 3rd degree Black belt in Karate.

OPEN SOURCE

Leave a Reply

(required)

(required)


ADVERTISEMENT

Gartner Infrastructure


Gartner Application Strategies


IBC 2017

ITBriefcase Comparison Report