Search This Blog

Human Cognition vs Age into an AI

Imagine building an AI that spans a lifetime of cognition by training the AI with people ages 5,10,15 and so on up to 105, then combine the AI from each age group, all 21 of them, into a single AI using a more advance neurosynaptic chip at 10w or greater.

Each age group (every 5 years) would train their representative AI engine by answering a list of questions about sound clips, videos, images, concepts, ideas, language, thinking exercises, games, morals, ethics, and above all other category knowledge, extracted from Wikipedia and other sources like Wolfram Alpha :) Think of this as a carefully cultivated system of ideas, experiences, knowledge, information, skills, and big data.

These agents, all 21 of them would then communicate with and train a master AI with far more processing power, memory, and storage capacity for indexing knowledge bases, one based on a server sized neurosynaptic processing computer! Woot Woot IBM! Add in some Intel Micron Instant On X-point drives, and shebam, everyone can afford Jarvis from IronMan!

The human brain has limits, it is made of a finite number of neurons, billions, and connected with trillions of synapses. In the sciences these numbers are not that big. In computer science today for example it is not uncommon to have a Terabyte of storage in the hard drive. My late 2015 iMac 5K display has a 2 TB drive made of an 128gb SSD and a 2TB spinning disk called a Fusion Drive (c) Apple 2014. Named Fusion because it combines the speed of the SSD with the robust data capacity of a spinning disk in a virtualization drive volume managed by OSX in an intelligent way to reduce the time to needed to turn the computer on with the solid state drive speed, while automatically moving recently underutilized data to the big slower spinning disk. This way the OS launches quickly, without causing a penalty in cost associated with a 1 TB SSD ($700 via apple configuration for that iMac). You get more for you money, performance where it counts, and a better lower cost data solution with the 2TB and 3TB drive, sadly to reduce costs, apple went with a comically small 24GB SSD in the 1TB fusion drive configuration, avoid it, unless the lower cost is absolutely needed.

Back to the storage limits of the human brain, roughly 700 terabytes. Sounds big to computer people today, especially for a home desktop, but that is the approximate average index volume capacity of an adult human brain for ideas, images, memories, concepts, so on, all the things that make up cognition and awareness, consciousnesses, amazing natural intelligence that makes up he human mind. Our perspectives and ideologies underpinning our values, ethics, virtues and capacities of cognition. Love a complex mixture of virtues and ideology with feelings so deep that love makes everyone feel something profound, something computers and AI may never be able to perfectly duplicate, if not to mirror and impersonate beyond questioning!

Today the most advanced AI you interact with is googles search engine itself, more specifically its 30th generation bot crawler, with the brawn of millions of watts of computer power behind it, perhaps even billions of watts worldwide. Siri at Apple is state of the art, the natural language processing perfectly imperfect verbal search engine enabled on all new iPhones, and many older models as of 2016. When you call a large corporation you are likely to encounter an AI on the phone. We can see examples of advanced AI even in some toys like my Segway Ninebot scooter, or the Furbee. When the Furbee talks with Apple's Siri, it sounds like HAL from 2001 a Space Odyssey by Stanley Kubrick. The AI engines of today are amazing when you stop to think about how they work and what they are doing when you perform a voice search!

Back in 2010 when Steve Woz of Apple was asked about the future of computer science, he said that natural voice was going to be huge. Today many companies are working on natural language processing that is used as the front end of AI' that people encounter when communicating with large corporations all the time. Siri, Google, Cortana, "Ask me anything" is microsofts prompt for the AI named Cortana, inspired by the game series Halo, Cortana is a futuristic humanoid machine hosted female AI in the game series, one that works closely with Master Chief to help him achieve his objectives, the Xbox player immersed in interactive content that goes beyond the flat passive images of TV to something more interactive, deeper and more engaging.

Online lets the gloves come off, you get to hear what people really think in the comments on the threads. Today, Prism at the CIA is cataloging all of those comments, every phone call, every email, all big data. They take this data and feed it to AI agents who look for bad people, and every day the Prism program becomes better able to identify America's enemies both domestic and abroad. I would like to give props and thanks to the IT staff of the CIA, those who funded Prism and all of those who work on making it all happen to keep America safe. Here is the only example of futuristic AI that we even know about! Good work Government! I have to applaud the US government for this one. I am often not proud of my governments net overall effectiveness at representing the public, nor impressed by its operation efficiency, but I am very pleased with the Prism program, and other secretive programs designed to give America the edge over the next century. I would like to thank my friends in Congress and the Senate for supporting advanced technologies, especially the compact fusion reactors at General Atomics. In the future compact commercial fusion power will power almost everything, more on that later. For now rest assured that the end of fossil fuels will not end humanities ability to mimic the stars above, space based fusion reactor stars like our sun, from which science, all human knowledge building reverse engineered an energy solution from the sun, based on physics and chemistry happening in the sun, presumably, or on its surface where we can see it more clearly with special imaging equipment, space telescopes, and a whole range of very expensive custom made imaging sensors that span a large portion of the electrodynamic spectrum!

Perspective makes all the difference, people today uses optics, telescopes, electron microscopes, lidar, CMOS sensors, CCD sensors, and elaborate thermal imaging sensors and other wide spectrum x ray imaging sensors, deep IR sensors, and many other machine vision IO technologies to see like a MRI, CAT scanner, and all the other ways the people can see using special machines today. These are the potential eyes for an AI system. The images available online are another body of imaging data that computer AI's can use a big data library from which to understand the world, with intelligence.

The Neurosynaptic chip was designed to mirror the architecture of the energy efficient human brain. This means that a 2.5w neurosynaptic chip can do more computational data processing then a 120 van neuman archtecture i7  chip from Intel, vastly more than a small server farm made of Xenon processors, and a very small neurosynaptic chip optimized for machine learning can learn like a 5 year old human today, hosted in a low power device like a tablet today, but are at the cutting edge of technology and not exactly something you will see at teh 2017 CES or on sale at Best Buy anytime soon.

We are already building advanced AI technology as a civilization, so lets all get behind it. It will make the future safer and more secure, reducing crime and corruption, it will increase transparency across all of society with ubiquitous universal online surveillance networks :) Telecom and big data have whole new industries ahead of that cloud hosted video stream Drop cam action I have been using at home for about 6 months :)

No comments:

Post a Comment