Friday, January 13, 2017

Building Human-Level A.I. Will Require Billions of People

The Great AI hunger appears poised to quickly replace and then exceed the income flows it has been eliminating. If we follow the money, we can confidently expect millions, then billions of machine-learning support roles to emerge in the very near-term, majorly limiting if not reversing widespread technological unemployment.

Human-directed machine learning has emerged as the dominant process for the creation of Weak AI such as language translation, computer vision, search, drug discovery and logistics management. Increasingly, it appears Strong AI, aka AGI or "human-level" AI, will be achieved by bootstrapping machine learning at scale, which will require billions of humans in-the-loop


How does human-in the-loop machine learning work? The process of training a neural net to do something useful, say the ability to confidently determine whether a photo has been taken indoors or outside, requires feeding it input content, in this case thousands of different photographs, allowing it to generate its own model of the photographs, correcting, re-generating and improving the model until the program has achieved a high enough confidence to perform the sorting behavior automatically. This neural model can then be applied to other content and ultimately requires less correction by humans in the future. Thus, work has been done and added to the broader body of machine learning knowledge.

One can then imagine that, over time, as these models are encoded, fewer and fewer humans will be needed to train up useful AI... Wrong!

Rather, as the companies now trailblazing AI (Google, Amazon, Apple, Microsoft, Facebook, Tesla, Uber, etc) have generated more value through machine learning, they've realized that 1) machine learning can be applied to infinitely more domains/problems, 2) that more complex, creative problems require more human-in-the-loop intervention, and 3) that more value can be created by integrating the machine learning they've already done - a cumulative effect, eg Google's recent breakthrough in translation, which ultimately required billions or trillions of human-in-the-loop (including you, if you ever used Google Translate) machine learning cycles to finally break through to another level of automatic functionality. 

To recap, machine learning requires 1) some well-educated machine learning professionals, 2) many more less-educated machine learning guides and 3) access to large swaths of structured content, 4) access to previously encoded machine learning. And the market-driven desire to apply it to new problems sets is growing very, very quickly. 

With technological unemployment growing as a U.S. and global problem, and economic stratification rapidly increasing, many have been wondering how the general human population will earn a living in the transformed economy. A few years ago I argued that users of social networks could soon start getting paid by the parent companies. Now that a the basic business model surrounding human-directed machine learning, AI and digitized content is emerging, that scenario can be advanced. 

As the Great AI Race heats up and more companies, countries and other actors come to realize the narrow and broader potential of human-in-the-loop machine learning, the demand for machine learning pros, machine learning guides and content workers will grow proportionately, driving up their share of the pie as they help to build more intelligent superstructures brick by brick.


The growing competition is also driving up the value of content itself - especially large bodies of structured content. Over time, content producers (including users of search engines and social networks who add value simply through their interactions with those systems) can expect to receive more value for their work or property.

As AI-generated revenues continue to grow, additional billions, even trillions of dollars will flow to super-lucrative machine learning processes and, ultimately, into the digital pockets of the masses essential to building the different aspects of AI. 

The amount of value shared with users will depend on the size of the pie. With Kurzweil's Law of Accelerating Returns in full effect, that pie is likely to grow MASSIVELY. The limits to growth appear to be our finite ability to capture, sort and export information about our lives and the universe around us. In theory, the total pie is limited only by the total information contained in our universe. 

From one perspective, this process can be viewed as a market-driven acceleration of science. From another, it's an evolution of the economy from Industrial Age to Knowledge Age. Looking at the big picture, it sure looks like mass-scale Human/AI symbiosis that ultimately drives up machine, human and planetary intelligence by digitizing the vast universe of information surrounding us.

Seems pretty natural to me.

Saturday, December 17, 2016

Donald Trump, Entertainer-in-Chief

The days of the presidential presidency are behind us.  

JFK was the first TV President. He and his successors exuded a distinctly presidential vibe as they communicated confidently to the masses, primarily through color video, usually behind a podium or in high-power settings, on a monthly or sometimes weekly basis.

Donald Trump is the first Web & Reality TV President. He spent a decade as host and producer of the hit show The Apprentice and exudes a distinctly colloquial vibe across cable and the web. Trump prefers titanic business settings like board rooms and communicates to the masses at a daily or even hourly rate, even after the election. Twitter is his pulpit.

Trump is a seasoned, self-aware, master content producer AND actor. In sports, the equivalent is a player/coach, a Peyton Manning or LeBron. He's calculatedly sloppy and unpredictable, which appears to boost his authenticity and watchability. Most importantly, he's relentless.

Trump's magic media formula: 

  • Trump is entertaining and highly watchable. He understands how to produce good drama, keeps messages simple and repetitively hammers home points he wants to get through noise. His content gets people talking, furthering his reach.
  • Trump uses social media effectively. He's mastered Twitter, Facebook and YouTube and uses these more intuitively, thanks to years spent in entertainment, furthering his reach.
  • Trump producers content at a high volume. YouTube producers and social media marketers know that if you stop feeding your audience, even for just a week, they will disappear. Above all else, Trump understands how to maintain and grow a captive audience and maximize reach. He can saturate mass media to his benefit and the detriment of other signal.

Like him or not, President-Elect Donald Trump is a highly entertaining, dramatic, calculating, content and social signal master with skills and a team tailored for contemporary media. His election victory, not unlike JFK's, can be attributed, in part, to a modernized communication playbook (which may also be at work in places like Russia and the Phillipines) and marks a notable shift in American sociopolitical landscape. 

Donald Trump may be the second Actor-in-Chief, but he's definitely the first Producer, or more fittingly, Entertainer-in-Chief. And he's got the spotlight... for now.


Thursday, February 18, 2016

IBM Watson AI XPrize Pits AI vs. Human/AI Teams


XPRize and IBM have announced the IBM Watson AI XPRIZE, a multi-stage Cognitive Computing Competition with a $5 million purse that challenges "teams from around the world to develop and demonstrate how humans can collaborate with powerful cognitive technologies to tackle some of the world’s grand challenges." Interestingly, the competition will be open to human/AI hybrid and exclusively AI entrants alike. The contest will culminate in 2020 after a series of IBM's annual "World of Watson" prelim events and draw attention to the human-empowering aspects of Artificial Intelligence. 

May the smartest neural array carry the day.

Pre-registration is open now at xprize.org/AI, and detailed guidelines will be announced on May 15, 2016.



Ukraine to Test Transparent, Blockchain-powered Voting System


A handful of Ukrainian government officials, several Ukrainian NGOs and blockchain businesses Ambisafe, Distributed Lab and Kitsoft have announced a partnership to test and gradually implement a transparent, blockchain-based voting system in Ukraine. The move to the blockchain marks a next logical step in Ukraine's recent experiments with e-voting systems.



Tuesday, November 5, 2013

Ingress - A Precursor of the World to Come

With over 500K active players, Ingress, the Google-funded augmented reality game for Android, is about to exit Beta and already marks a notable step forward in gaming and interactive media. As it scales, it could have a major psychological and material impact on our world.



Ingress as Indicator: Futurists, tech bloggers, entrepreneurs, investors and sci-fi writers all spend much time scouring the world for interesting signals from the edge to identify emerging trends or even the next big thing. In the past decade, many have zeroed in on gamification, augmented reality and the ongoing mobile explosion as important zones of development. Residing squarely at the intersection of these potent growth areas is Ingress, the quirky augmented reality game that hearkens to visions of the future contained in works like Snow Crash, Otherland and Rainbows End. As I’ve played the game (I’m up to Level 7 of 8), I’ve come to believe that it’s an important precursor of things to come.


Gameplay: Created by Niantic Labs, an internal Google startup that’s technically autonomous from the parent company, Ingress is both simple and very complex. Fundamentally, it splits players into two teams, Resistance & Enlightenment, that play in the context of a sci-fi storyline that progresses week-to-week. The two massive sides battle, in real-time, to control and link portals located at real-life places of interest such as libraries, museums, post offices, restaurants, murals, etc. You play by walking or driving around to portals and capturing them, then linking the portals together into triangles called control fields. Whichever team has the highest # of human population contained beneath their control fields is considered the current global leader. Notably, players can also earn points for submitting new portal by providing GPS tagged location photos.


Flying Under the Radar: Boasting a large, highly engaged player base, Ingress is a fascinating case study on multiple levels. But it’s been able to fly under the radar largely due to the popular perception that it’s purely a game. Expect that perception to change as the Ingress platform continues to evolve and bloggers and other experts continue to notice its traction and potential impact on information gathering, work and social interaction.


10 reasons to pay attention to Google’s great Ingress experiment



  1. The Hanke Factor: Ingress is the brainchild of John Hanke, the creator of Keyhole, which was purchased by Google and then became Google Earth. Hanke went on to work as Google’s VP of Product Management overseeing the impressive Geo division (Google Earth, Maps, Places, Local, StreetView, SketchUp, and Panoramio) and then created Niantic Labs as an autonomous entity under the Google umbrella. He’s a big-thinker and achiever with a great feel for launching geo-products and platforms utilized by billions who is unlikely to waste his time on something not-massive. In light of CEO Larry Page’s focus on projects with massive scaling potential and his reported desire to keep Hanke and Niantic in-house, it’s obvious that Google views Ingress as more than just a game. Among other things, it’s an attempt to create a new mobile-phone-based system that can extend Google’s information gathering abilities into the rapidly emerging geosocial sphere.


  1. Serious Traction: In just under a year of closed beta 500k+ users are actively playing Ingress. I was lucky to receive an early invite and have played on-and-off since January. I very much enjoy the game, though many of my less techy friends regard it as big time-suck. Much like foursquare, it’s mostly appealing to geeks, gamers and early adopters. What sets Ingress apart is its well-designed gameplay, competitive structure and compelling secondary world & storyline. With some basic tweaks, it could relatively easily become an informational gusher for Google Places, which should make Yelp nervous and Foursquare incredibly so.

  1. Augmented Engagement: To date, there are very few examples of augmented or mediated reality games capable of garnering user participation and changing their real-world behavior. Geo-caching has been around for decades, but is a lone-wolf game with limited adoption. Foursquare was a breakout success, but appears to have stalled somewhat. Considering the level of participation required, cooperation required and sheer # of active players (who spend a lot of time coordinating with one another and traveling from portal to portal), Ingress, to my mind, is the most successful augmented reality game on Earth to date. This augmented engagement accomplishment is an important benchmark for a new class of technology that will likely power or catalyze new work and entertainment behaviors. Naturally it’s Google that has perceived this potential value and invested heavily. Mix in rapidly evolving mobile devices such as Google Glass, which Hanke and the Ingress team are already messing around with, and the feasible near-term future of engaged augmented gaming gets very interesting.

  1. Next-Gen Gamification: The idea that software-mediated games might serve as the interface for many future systems, ranging from social media to ordinary task management, caught fire in recent years. Made possible by powerful smartphones, growing bandwidth and better backend capabilities, the underlying Ingress platform represents the bleeding-edge of gamification. Some potent future capabilities can be extrapolated from a seemingly innocuous portion of the game that rewards players for submitting GPS tagged photos of interesting locations to become new portals. It’s no stretch to imagine that Google could apply the Ingress backbone to expand its Street View capabilities. Considering that Ingress players are now helping Google to quantify places of interest for free, imagine how these crowd-sourced mapping efforts will be amplified as cameras improve, bandwidth grows, new sensors are added to mobile phones (such as a miniaturized Matterport or Bublcam device), augmented reality gets real, gameplay gets better and financial and/or other participation incentives are increased.

  1. RL Social Interaction: A popular and valid criticism of social media and video games is that they fuel anti-social, glued-to-the-screen behavior. Contrasted with the likes of Facebook and Call of Duty, my time playing Ingress actually rewards me for moving about the real world and interacting with other players. My girlfriend and I enjoy playing the game together and find that it can increase our weekly physical exercise (which the game actually measures). It’s refreshing and illuminating to see technology making that possible. At the same time, the flipside of increased gas and bandwidth usage is a valid new age concern and interesting edge indicator.

  1. Media Component: Ingress utilizes a complex sci-fi storyline to structure its gameplay. Weekly video updates are released by in-game “reporters”. Factions work together in the context of this information to score more points and gain prestige. It’s hard to tell what sort of an impact this has had on engagement, but my guess is that it’s very important for super users and also helps retain general users. Seeing rich media storylines in video games and now in augmented reality games suggests that rich narrative structures and secondary worlds will play an important role in the ongoing roll-out of social media and the metaverse.

  1. Future of Advertising: Having worked on a pair of augmented reality startups (in3d, a 3d modeling and video company, and Swarmado, a content harvesting game for groups at events) I appreciate the importance of an underlying financial model that can support the software, server and human superstructure required by these ventures. By-and-large, advertising is the go-to strategy (though a few choice companies can bypass this if purchased for their sheer informational value or future potential). The holy grail of location-based advertising is generally considered to be any system that can meaningfully drive foot traffic to locations where its currently not. In this regard, Ingress could be a smashing success. Through partnerships with Jamba Juice and Zipcar, Niantic is actively exploring this option. I’ll be following new developments closely to see what strategies emerge and how closely they resemble or depart from those we hatched at in3d and Swarmado. :) There certainly are many potentially lucrative options for any system with a large user base and high engagement.

  1. Privacy Concerns:  Privacy concerns can be tinderboxes that cause unwelcome flare-ups for info-oriented companies like Google. Streetview has already drawn serious market backlashes in various regions such as Germany. As Google inevitably seeks to input more information about the world, the company carefully considers the social and political implications of its behavior. Though Hanke has stated his preference for start-up culture as the reason for Niantic Labs’ separation from parent Google, we should also consider the possibility that Page and Hanke both want to ensure the public and governments view the two as separate entities so as not to add to Google’s image as a great datavore. Ingress has already taken some heat for its info ingestion from bloggers and in forums, so that looks to have been a good call. As the project moves forward and the game evolves it could become a critical, perhaps revolutionary, tool for systematically gathering information about the world via swarms. It could become a mini-Google.

  1. Human Resources: 1) Imagine if Ingress, or a version of Ingress based on the same principles or backbone, established itself as a real-deal information generator. Then, 2) imagine if it became a formal part of the broader Google system. Such a gaming system could incrementally reward millions of player-workers for the data they input or even serve as a soft “play-in” layer for vetting future Google employees. Utilizing game-based systems for HR purposes could become increasingly effective and popular in the coming decades.

  1. The Google Factor: As listed above, there are many reasons for Google’s interest in a social, info-valuable augmented reality gaming system like Ingress. It’s even conceivable that such a company could pose a future threat to the search behemoth. To my mind, Google’s interest in Ingress reinforces its solid strategic vision of the near-future and that the company is making very smart bets on technologies that could both threaten or amplify its core potential. As a part of Google, Ingress could add billions to its market cap. Independent, or as part of a competitor, the quirky game might have become a major, central threat to Google.

Now that Ingress is moving out of Beta (it no longer requires an invite, you can get your Android download here) and onto iOS in 2014, you can confidently expect to catch many more smartphone-wielders to silently, or not so silently, playing the game all around you. Whether you find it to be a gigantic waste of time, gas and bandwidth, or a revolutionary new economic system, it’s an edge indicator that’s at least bound to garner hype and headlines for years to come. Seriously.

Friday, April 13, 2012

The World's First Billion Dollar Brain



How much might the highest bidder pay for Steve Jobs' intact brain in a jar? I could see a die-hard collector and history fan dropping 10 or 20 million $ for bragging rights. Maybe more.

But what if that buyer could count on extracting information from the brain? As science continues to better our understanding of functions like memory, intelligence and cognition, and improves brain-scanning and simulation, we're rapidly developing the ability to identify where and how information resides in brains. Researchers have already  distinguished between different recalled memories in brains. So how many more years will pass before mankind can read meaningful portions of the well-preserved brain of a deceased person? 5 years? 10? 20? 50? 100?  

The answer may well be 10-20 years, but even if it's 100 - that could seriously affect the price a person or an organization is willing to pay for a brain. The prospect of retro-active brain reading will surely push up the going rate for preserved brains.

So maybe that raises the top bid for Steve's brain to $50 million, especially considering the advances being made in fields like chemo-preservation, which is turning out to be a powerful alternative to freezing. For an in depth overview check out Ken Hayworth of the Brain Preservation Foundation discuss this at length - begins at 3 mins 30 seconds.


Keeping in mind accelerating developments in computer processing, scanning, information theory and brain sciences, how much might a government or group betting on the prospect of retro-active brain reading in the future be willing to pay for the brain of a key scientist, intelligence operative, general, politician, inventor or enemy? The going rate will of course be influenced by available budget and the certainty of the purchasers, but both of those will only rise as we move forward in time. There will most likely be more capital available. There will most likely be greater certainty that brains can be read. 

How much would a company be willing to pay for Steve Jobs' brain (which was not preserved, btw)? How much would the U.S. pay for the intact frozen brain of China's leading cyber strategist? How much would Obama's or Gates' chemo-preserved brain be worth to the Chinese govt? 

The only thing I'm really sure of is that the informational value and going $ rate for ALL brains will rise steadily over time as accelerating change, systems quantification and the emerging superfluid economy conspire to allow us to read brains better and make information more useful and transferable - just as the value of rain forests goes up as we figure out the previously discounted or externalized value of the ecosystem and species therein.

Perhaps we'll get to a billion dollar brain purchase by 2030 or 2040. Perhaps the average going rate for a brain will be $1 billion (inflation adjusted) by 2060. Perhaps some foresighted risk-takers are already stock-piling brains. Perhaps there are companies or nations that have already put "brain recovery" clauses in key asset's contracts. 

It may seem crazy to contemplate. But such an exercise is much less whacked than it would've seemed just 5 or 10 years ago. 

The main point: it's worth thinking about - maybe someone will retro-actively read your brain one day.

Monday, April 9, 2012

The Massively Multiplying Mini Me

The notion of the Quantified Self has been gaining popularity as people figure out more useful things to do with the data captured via their computers, devices and social networks. Many thinkers and companies (like fitbit) imagine this data can and will be used as a tool for revealing personal heath trends in areas like sleep, exercise, happiness, bodily functions and genetic disease. Some folks like mathematician Stephen Wolfram, who already are mining vast stores of personal data from over the years, believe introducing the appropriate search algorithms and systems to personal data will help us with identifying behavioral tendencies and thus help us be more productive. My friend, ASF President John Smart, takes a bigger more forward-looking perspective, sees a world in which this data will be used to create a Digital Twin for each of us that will fully anticipate our preferences and behavior and greatly assist us in all aspects of life.

I think they're all right.

That said, I believe there are also many less useful, but more entertaining uses for these growing pockets of personal data. Set up correctly, these could not only scale quickly, but also help people to more quickly realize the value of their personal data.

So here's a near-term Quantified Self product idea to test that assumption: the Mini Me or Pet Me.

Imagine taking all of your gmail, facebook, computer and sensor data, then putting it into a simple 3d avatar that looks like you and sits in a window on your facebook page or anywhere else on the web. Perhaps this Pet You is displayed on a wall in your house. You can interact with this Pet You - ask it questions, push it around, introduce unexpected elements like warm apple pie or a swarm of bees. 

Its responses all draw on your personal data history. Maybe 1 in 5 responses is really funny or bizarre. You yourself, family, friends or other users can then rate each interaction to guide the development of this Mini You, essentially raising it by rewarding the most desired behavior. (Maybe this growing up portion of the software is called Skinner Box.) 

Over time, some of these growing Mini Yous are bound to get interesting. These are the ones that will literally survive and thrive by being shared on different people's facebook pages, via email, in public settings and... this is where it can get really interesting ... as Non Player Characters (NPCs) in bigger virtual environments like Second Life, single player big world video games like Skyrim & Assassin's Creed, as cleverer A.I. in first person multiplayer games like Call of Duty & Halo, or as MMOGs like The Sims Online.  

Currently, leading edge video games utilize flavorful, but simple NPCs based on personality archetypes and solid writing. There's amazing craft that goes into constructing massive storylines and the corresponding character traits, language and behaviors. But NPCs are poised to very quickly grow in complexity as social and quantified self data is introduced into their systems. It occurs to me that whichever company can encourage people to grow the Mini You en mass will create a new category of NPC Unit or Tiny A.I. that can easily be introduced into a wide variety of these games and worlds to beef up, augment or replace these existing NPCs. 



I bet that over the next few years these little Mini Mes will start multiplying - initially driven by startups focused on scaling them across social networks. The first few gaming, app or super-simple A.I. companies to create interactive Mini You farm systems that are engaging and entertaining will probably claim dominant positions (just like Foursquare and Instagram did - scale first, get more complex later). It probably won't be the Wolframs, the Googles or the Hard A.I. start-ups that grab the initial user base and the big funding. It'll be some scrappy little team that builds a Mini Me game that's even more fun than Farmville, Angry Birds or even the current king of the castle Draw Something.

Then the entrenched right-fit entertainment and communication power houses will scramble to introduce these Mini Yous across their products and services, thus fueling the diffusion fire.

The result? As we use our personal data to grow our little Insulting Irmas, Cowboy Karls, Angry Alvises and Witty Williams we'll not only laugh, be shocked and express outrage - we'll also begin to understand what it means to embed our patterns into popular video games, into highly customized video games, into advertising, into the increasingly complex and pervasive web - and quite probably start earning money or other social credits for these forays. Many unexpected things will occur as these Mini Mes escape, like the dinosaurs of Jurassic Park, and begin to live autonomous lives as temes, co-mingling with viruses, surviving and evolving wherever there is silicon to be found, changing the nature of the web.

At some point the Mini Mes will surely match the capabilities of full on Digital Twins, but expect a lot of product iterations and cycles and evolutionary branching before then. Like life itself, these authorized and unauthorized little genies will resist all sorts of bottling, adding increasingly more simulations of ourselves to the wild west of the web.