5/25/2018

Smart Machines Impact on Society


        I.            Rise of Smart Machine

Smart machines pose a threat to society.  Smart machines are simply algorithms that are self-taught.  They learn by processing mountains of data as input, learn patterns over time, and respond accordingly.  The notion of “Artificial General Intelligence” or AGI for short, is the goal for smart machines.  Many organizations see the potential to create AGI for a variety of reasons.  These organizations are well funded and have access to the latest technology, variety of information and industry leaders.

      II.            Easy Access to Technology

Computers can crunch numbers much faster than humans.  Computers interpret inputs in a variety of formats, the code executes a series of instructions and complies accordingly.  Computer algorithms continue to advance, hardware and software prices continue to decrease and increase in availability to larger data set the stage for anyone to dabble in smart computing.

    III.            Surpass Human Intellect

Computer programs can process huge amounts of data, learn patterns over time, and respond accordingly.  Over time, smart machines can and will outperform humans exponentially in a variety of tasks.  Likewise, smart machines with specific domain knowledge could be integrated with other smart machines with specific knowledge in other domains.  This could create a network of smart machines with tremendous information.  That network could gather information and respond according in real time.

   IV.            Accumulation of Data

o   Larger data sets are proliferating
o   Both Proprietary and Public Domain

     V.            Labeling the data

o   “Supervised” Machine Learning algorithms require manual data classification
o   “Unsupervised” Machine Learning algorithms do not require labeled data

   VI.            Algorithms Aren’t “Teal Time”

o   Takes time to move, prepare and ingest data
o   Training the data models takes time

 VII.            Data Models

o   Models are statistical probabilities and not designed to be 100% accurate
o   Domains are typically isolation and don’t “communicate” with other domains yet

VIII.            Shortage of Data Scientist / AI Engineers

o   Although Universities are sending qualified technologist into the workforce, a shortage of Resources with AGI experience still exists
o   The basic process to create smart machines is complex and limited to some of the more advanced thinkers

    IX.            Black Boxes

o   Unfortunately, the underlying mechanics that allow computers to interpret raw data into useful information is a “Black Box”
o   The power “under the hood” of current algorithms is not completely understood

      X.            Ethics and Bias

Another area of concern are the underlying ethics behind the algorithms.  There is potential to add “human bias” into the smart machine software, by design or inherent in the human process.  Those bias may be minor, yet they could be amplified with serious downstream consequences. 
·         Who gets to decide which rules are acceptable and which aren’t?
·         Can rules be modified over time? 
·         Is there a governing board that ensures the smart machines behave within acceptable behavior and what are the ramifications if they deviate from the stated objective?
·         If smart machines do something outside the bounds of acceptable norms, can the owner of the software or robotics be held accountable?
·         If AGI is created outside traditional incubation organizations, how can we guarantee rules are followed?
·         How can we ensure the smart system has an “off” button as precaution?
·         Who will monitor the network of smart machines and how will that be funded?
·         Will “laws” be created or modified to account for smart machines and robotics?
·         What “rights” will smart machines have?
·         Who “owns” the smart machines?
·         Do “Smart Machines” have rights?
·         Can “Smart Machines” and their designers be held liable?

    XI.            Bugs

Since the advent of computers in the middle of last century, bugs have appeared in nearly every software program.  Bugs are features of the system that were not accounted for or do not behave the way the developer intended.  For example, if you attempt to divide by zero in a calculation, that will result in a bug.  If you don’t account for that “bug” ahead of time, the program may behave in unintended ways.  It’s nearly impossible to remove all bugs from every program as many bugs can’t be predicted ahead of time.  Also, smart machines could eventually write their own programs which are more efficient at solving problems.  How can we ensure they write code that doesn’t contain bugs with unintended consequences?

  XII.            Quantum Computing

Quantum Computers are starting to make an impact in technology by leveraging some cutting-edge technology.  Although mostly large software companies, joint ventures, academia and space programs have availability of funds and technology, a newer version of computing has been developed.  Basically, it leverages the traditional “binary” mode of computers that use a combination of 1’s and 0’s and adds a third element of Zero, One or Both.  This allows computation of gigantic data sets and once insurmountable problems to be solved with increased speed.  The major downside is having the ability to solve problems of encrypt ability.  If a Quantum Computer can break the code of our most sophisticated algorithms, how will society encrypt things going forward and what alternative methodology will be used to ensure secrecy and accountability.

XIII.            Elimination of Jobs

We are witnessing some reduction of specific jobs in the workforce due to automation.  Some processes can be better performed by smart machines or robotics.  They contain fewer number of errors, more efficient, do not require much time off and can be recorded as assets on the financial statements that get depreciated over time.  This phenomenon will continue to increase into other niche jobs.  It could also expand into many segments of society which were “automation proof”.

XIV.            Unemployable Workers

As more jobs become obsolete through automation and smart machines, the number of unemployed individuals that never return to the workforce could increase.  How will these individuals sustain basic needs of food and housing?  From a financial perspective, will loan defaults increase?  Will crime increase?  How will the economy continue its upward growth trend if fewer people have access to goods and service and lower discretionary income.  Are there measures in place to ensure civic order?

 XV.            Basic Living Wage

There has been discussion of the Basic Living Wage to provide capital so every person can sustain life.  Although this concept makes sense, there are many holes in the model to support widespread unemployment.  Who will finance the money to support a good chunk of society.  How will people occupy themselves 24 hours 7 days a week?  Will the number of offspring be curbed to alleviate the burden on society?  How will the education system be altered if the graduates have limited chance to find gainful employment?

XVI.            Spiral

As organizations strive to increase profits and reduce costs, Smart Machines have the potential to automate many tasks currently supported by humans.  As more people get displaced from the workforce, few purchases of goods and services could result.  This translates to less monies flowing through the system.  This has potential to lower the tax base collected by institutions.  It could also amplify a need for further automation to maintain profit levels, resulting in further job loss.

XVII.            Conclusion

The Rise of Smart Machines creates many opportunities across a segment of society.  Yet it also poses some obstacles to overcome downstream.  Increases to technology have the potential to alter a good chunk of society and great minds need to come together to discuss some of the ramifications and plans for a soft landing.

XVIII.            Bibliography

         I.            Gartner Says Smart Machines Will Have Widespread and Deep Business Impact Through 2020
       II.            Accessibility
     III.            Superintelligence
    IV.            Responsible Conduct in Data Management
      V.            Training / Learning Methods
a.       Supervised Learning
                                                               i.      https://en.wikipedia.org/wiki/Supervised_learning
b.       Unsupervised Learning
                                                               i.      https://en.wikipedia.org/wiki/Unsupervised_learning
    VI.            Real Time Computing
   VII.            What No One Tells you about Real Time Machine Learning
 VIII.            McKinsey Report Highlights the Impending Data Scientist Shortage
     IX.            Black Box
       X.            Ethics of artificial intelligence
     XI.            AI expert Nick Bostrom: Machine super intelligence will have no off switch
   XII.            Quantum Computing
 XIII.            Automation and Anxiety
 XIV.            AI, Robotics, and the Future of Jobs
  XV.            Living Wage
 XVI.            Deindustrialization



Work Life Balance

This past Wednesday was 2 year anniversary at current employer.  I was let go from prior job, after 3 years or so, along with other developers, the work dried up.  I reached out to network and found the new job within a week.

The company I work for was acquired by a larger company.  Although I work out of the house mostly, my home office is in Indiana, and the main headquarters is in Maryland.  I worked a few contracts in Indianapolis, and been working for Boston client since March of 2017.

Although I work from home, for the past 5 years, I've been lucky enough to work for a variety of clients, in multiple cities and states, variety of projects using different technology, in multiple industries.

People often wonder, if remote workers keep busy while working from home.  I'd have to say, "definitely yes".  I'd say we get more work done actually.  There's no commute, so we can log in 5 am some days, get busy before everyone wakes up, and we work after hours more often than not.  Although work is performed during core business hours, there's built in flexibility, to allow for doctor appointments and such.  So long as the work gets done on time, everyone is happy.

No commute, always connected, communicate with Instant Messenger, Email, Cell Phone, it's almost the same as working down the hall.

There is cost savings on gas, commute time, dry cleaning, meals, snacks, coffee, etc., we have incurred costs such as internet, electricity and cell phones.  In my opinion, the trade off is well worth it.

It's not just corner office, salary, time off and benefits.  Flexibility of time is a definite factor in today's workforce.  From suite & tie & white collar starched shirts, to casual, to remote.  The progression has increased the flexibility options, resulting in higher performance.  Seems like a win-win for everyone!

And so it goes~!



5/22/2018

Can Robots Become Insane

Pardon me, but your robot is insane.  Say what?  It does not comply with the formal structure of reality.  How so?  It has complete disregard for social etiquette for starters, next, it disregards basic rules of the law, and by all means, has no manners what so ever?

It's possible for future robots to be considered insane.  For having a reality other than the accepted norms.  And for that, we can simply unplug, remove from society and / or update its programming.  That should suffice.

For a robot to become sane, it must comply with cultural norms.  We could program a model for each country, each city, so the robot would integrate seamlessly.  Perhaps the local could be purchased as an ad-on, or when visiting another city, purchase the upgrade temporarily, for a fee.

We do have documented laws, which could be electronic in nature, those could be integrated into a model as well, by country, or state or city, same thing, purchase as add on, maybe buy a bundle of the entire South East US states.

In our society, the basic definition of an "insane" individual is this, 2 questions, is he or she a threat to other people and is he or she a threat to themselves.  If they can pass those two questions, chances are, they are good to go.  So too, a polite robot, that keeps to itself, is of service to others, and does not harm anyone or anything, and complies with the encyclopedia of laws across the land, would in fact be deemed "sane".

Can a robot in the future interact with humans?  Sure, why not.  Interpret language(s), speech patterns using sound input device, manorisms using sight input device, perhaps it can maneuver, sit or stand, climb steps, get in and out of automobiles.  Robots could interact with society, in a non threatening way, to be of service to humans.

There is not a point in which a robot would have a need to reproduce, so gender is non-mandatory.  Without a need to find a mate, it would reduce a tremendous amount of issues that humans create.  Robots would not have possessions, so again, no need to steal material possessions, or money for that matter.  As you can see, robots would not be motivated by sex, greed, , money and power, therefore, much of the issues would not arise.  So long as robots stay within the law, are not a threat to humans or other robots, I'd say there's no reason a "sane" robot couldn't co-exist in what we consider "reality".

In our reality, people do break the laws.  They do lie.  And cheat.  And murder.  So in that sense, if a robot is to co-exist, it must understand what humans are capable of, so they are not surprised when these events occur.  And they must be programmed to not do such things.

On top of that, there are the fuzzy laws.  Like going 5 miles over the speed limit when no cars around.  Killing does happen, in war.  Sometimes people alter the truth, to protect the innocent.  These fuzzy area's are probably a tad more difficult for a robot to comprehend.  Some rules are well defined, while others are hazy.  How does a robot know to differentiate?

If you ever watched Star Wars, C3P0 was a droid, that got along well with others.  However, he was purchased and sold, so droids are considered property.  Surely, they don't have equal status to humans in the movie.  And when his power is shut down, he goes to sleep.  And he is outdated as far as models go, so perhaps his purchase price is less than a current model with more features.  Does C3P0 know he's a droid?  In any of the movies, did he ever vote?  What rights did he have actually.  R2D2 seems to roll around, even in the desert sand, he was a bit shorter, and didn't talk much, just beeped on occasion.  Yet he had virtual reality ability, so that was cool.  They both interacted and integrated into mainstream society.  On different planets and cultures, and seemed to adapt okay.

So even in science fiction from the 1970's movie, robots blended well with humans.  I don't think the droids had consciousness or a soul, but perhaps that's open for debate.  Regardless, society has a place for machine like robots, when well behaved, or "sane" they might just fit right in.  And humans were not sent out to pasture, as we sometimes predict in our blogs and news articles, me included.

And there you have it~!


5/21/2018

Recent AI Article Summary

https://www.theatlantic.com/technology/archive/2018/05/machine-learning-is-stuck-on-asking-why/560675/?single_page=trueI read a good article recently.  Talking about the current state of Artificial Intelligence.  AI has made drastic improvements over the past few decades.  In that it can predict based on probability, as in curve fitting mainly.  For AI to make leaps and bounds, the intelligent machines must learn to "Infer", as in forward and backward reasoning.  In other words, give a result, have the machine determine the root cause, as well as predict the cause, based on initial data.  AI machines need to determine the reason "why", through use of connected AI systems, through communication, based on the environmental conditions it gathers over time.  Basically, AI needs to understand "reality", and then predict future as well as past causes.  

Here's the article link: How a Pioneer of Machine Learning Became One of Its Sharpest Critics

I don't claim to program AI systems, I enjoy keeping up with latest trends.  And surely AI is a current trend.

5/17/2018

3 Hot Tech Trends to Disrupt Everything

Artificial Intelligence on the edge.  Latest cutting edge technology.  In other words, let the AI model reside in Internet of Things devices sensors.  Ingest data, ping the Model, look for anomaly, fire off message to home base to alert.  Models can be created locally, pushed to the edge, where they reside.  Can update Models over time.  Seems like a good distributed AI model in real time.

We have devices to monitor people's vitals in real time, send messages back to home base, to alert if need be.  Combine the two, you've got some serious monitoring ability.

Due to security concerns, people have suggested embedding "chips" into children, so they are easily track-able.  Some push back from advocates, borders on ethical concerns, do we want to cross the boarder on people basic freedoms.

We can embed chips in cows perhaps, monitor them from a distance.  Seems like a stone's throw away, humans could be next.  First people volunteer, then offer service, similar to Flu Shots.

Internet of Things had security concerns out of the gate, opens up vulnerabilities, someone could tap into your home security through your thermostat, once in, scan your files, embed Trojan Horse, even Ransomware.  These are not good, yet they are real threats.  Suffice to say, if someone wants to hack you, they can usually find a way, as any device connected to internet is suspect.

With talks of cyber currency to overtake traditional paper money, we could soon see the disappearance of physical wallets.  Then all transactions will be documented in real time, audit trail, using the upcoming technology Blockchain.  Basically a distributed ledger system to handle transactions.  It uses a technique to add new transaction to the chain, by collectively validating the hash key, which is unique and created by hashing the prior key.  If you transaction is valid, it will be added to the stack and committed, and can never be altered, modified or deleted.  This should allow a valid history of all transaction.  With that said, financial transactions would no longer need to go through traditional methods, where the money is placed on hold until the nightly batch pushes the monies here and there.  It will be instant.  This can and will be applied to currency, voting, stock transfers, healthcare records, just about anything and everything.  This will disrupt all industries.

Taking the IoT example, what if they add microphones and cameras to people.  That would surely open up new avenues for monitoring, instead of current methods of smart phones, smart listening devices and cameras littered throughout society.  It would be a tighter mechanism for sure.

So it would appear, the latest hot trends in technology surround Artificial Intelligence, on the Edge, using Internet of Things, along with Blockchain.  These three technologies are primed to disrupt everything.