AI Will Be Making Decisions For You
AI will be making decisions for you.
Published November 17, 2015
Many consider Yann LeCun as the father of machine based OCR- the forerunner of the pattern recognition technology at the core of AI based Deep Learning. After occupying prestigious positions at Bell Laboratories, AT&T and NYU, LeCun was hired by Facebook in 2013 to head up their AI Research division know as FAIR – Facebook Artificial Intelligence Research. Facebook, along with tech giants like Microsoft, IBM and Google are sparing no expense in this area as new and enhanced versions of AI have proven superior at picking up skills to better challenge machine deep learning processes than their human counter part engineers.
The engineering and programming side of this challenge has always been centered around devising machine based neural networks that mimic biological based brain functions by building intricate replicas of the brain using artificial bio systems to construct inorganic based neural networks. That challenge appears to have been met.
Numenta, a silicon valley start up is focused on understanding and exploiting principles behind now human wet ware works to replicate it in machines running AGI.
Vast data bases or data dumps are required in order to teach AI to learn in a manner similar to the human mind through a machine reasoning process known as Deep Learning. Deep Learning, unlike mathematical based algorithms, is dependent on a process of pattern recognition, a biological function performed by the human brain. AI software engineers have been baffled regarding the roles played by the thousands of synapses connected to each biological neuron in the overall human reasoning processes which change and adapt as learning progresses by way of a type of electrical pattern recognition……….until now.
AI software engineers at Numenta have discovered that proximal and distal synapses actually perform different functions and that the distal synapses act as pattern differentiators in the thousands of electrical signals or patterns being transmitted on a biological neural network. This allows the synapses to not only recognize the presence of a particular electrical pattern, but to predict or learn the sequence in which the pattern will appear. This provides not only the basis for brain learning but also the basis for learning recall.
Engineers have also determined that it is not electrical patterns themselves that are key to rapid identification, but the differences that are actually key to the patterns themselves.
For example, the information content required by the human brain to recognize a word, which can have images and associations attached to it, is much grater than the information content associated with a number which say has no image or association connected to it, as in an of itself, it’s value is finite.
The process by which the human brain utilized thousands of synapses connected by even more proximal and distal dendrites has been an obstacle to creating AI machines to mimic the learning process of biological brains using pattern recognition. The pursuit is on to create artificial neurons with single point processing capabilities vs. the distributed, decentralized processing model employed by natural biological wet ware.
LeCun is aiming for something much more sophisticated than his pattern recognition software for AI. At Facebook, he’s pioneering AGI Deep Learning technology that will provide both the ‘common sense’ and communication skills necessary to converse with machines using voice alone, whereby the machines are capable of understanding sentences, express communication clarification and make suggestions and recommendations to it’s human interface.
LeCun goes on to explain that machines with these capabilities may eventually know what’s better for us than we do. He goes on to say, that these machines will not only be able to determine what entertains us and provide that stimuli, but it will also determine for us what we need to be exposed to, shown or taught, whether we deem that experience pleasant or not. He’s talking about creating a version of SIRI with a personality—-and an attitude.
In 2012, LeCun and Hinton [Facebook and Google] developed an enhanced version of visual recognition technology called ImageNet which was accurately able to identify and classify images at an astounding 85% accuracy by employing multiple neural layers on a single artificial neural pathway. This type of Deep Learning technology has now replaced the basic mathematical algorithmic core of AGI.
LeCun’s R & D department at Facebook is now putting the finishing touches on this technology to display what we call, ‘common sense’.
In order to perfect this, the systems would require access to vast sums of unrelated data – like we see being collected by the NSA and various other national security agencies and social media platforms. Facebook is already on top of this with the creation of a virtual ‘data butler’ called MoneyPenny to be released shortly.
MoneyPenny will assist Facebook and Googles cross platform Deep Learning AGI to make decisions for those human counterparts that interact with it. Both the launch date and technical specifications on MoneyPenny are being held close to the vest by Facebook, but one thing’s for sure – – it’s reach and scope will be far greater than it’s existing companion……messenger.com.
The process of melding man with machines forming a hybrid civilization is well underway. Google, like Facebook is retooling it’s AI core processes moving from algorithmic based AGI to Deep Learning technology to drive and ultimately provide a self evolving platform for it’s own version of AGI know as TensorFlow.
Not to be outdone by Facebook, Google recently purchased DeepMind Technologies in 2014….an AI startup, for $400 million – one of the largest tech acquisitions to date – an AI tech company that happened to be founded then sold by Elon Musk who seems to be heavily vested in AGI from a profitability standpoint, yet is very vocal, publicly, about issuing warnings regarding it’s inevitable potential to become a run away technology that would have devastating effects on humanity as we know it.
One last note. The 9 tech giants moving into developing AGI cores who are comprising and controlling the vast domain of the IOT [the Internet of Things] include:
IBM
Facebook
Skype
Google
Apple
LinkedIn
Microsoft
MetaMind
….and 3 of Elon Musk’s ventures, SpaceX, Vicarious and NeuroVigil
The paradigm shift occurring in machine intelligence is not only designed to benefit these tech giants in the form of advanced intuitive AGI and creating broader profit channels for their industries, but it is being designed to ultimately remove any and all decision making processes from the biological human being, to the all knowing, sentient machines. These advancements in AGI are designed to further narrow the gap and distinction between man and machines bringing humanity one step closer to transhumansim.
Research Links:
Single Artificial Neuron Taught to Recognize 100s of Patterns ~ Technology Review 11.12.2015
Teaching Machines to Understand Us [Yann LeCun] ~ MIT Tech Review 8.06.2015
AI Advances Make it Possible to Search, Shop with Images ~ Technology Review 11.17.2015
9 Tech Giants Investing in AI ~ Tech World 11.17.2015
Additional Info:
CONVERGENCE: The Day After AI Singularity
How Deep Learning Works and How Google Is Artificial General Intelligence On The IoT
DIGITAL SPYING | Who & How: THE GHOSTS IN THE MACHINES
GHOST IN THE MACHINE – Part 2: Big Brother Is Bigger Than You Think
NSA Project Bumblehive | Mass Surveillance & Data Collection on Everyone
AI will be making decisions for you. AI will be making decisions for you. AI will be making decisions for you. AI will be making decisions for you. AI will be making decisions for you.
