Friday, April 26, 2013

Your Best and Worst Frienemy: Computers

We all love our smartphones and super fast laptops (well, at least after they start up). And furthermore, we can all agree that we couldn't survive without them. Imagine losing your email, word docs, and most importantly, facebook.

But what if there was a darker side to all of this? What if these computer chips are taking over our jobs? Unfortunately, 60 Minutes seems to think that this is a possibility.

We have all heard about this threat. Back in the early 2000's everyone said that "In 10 years, we'll have self driving cars so truck drivers won't exist." The only problem is...that they were wrong and as a result, most people dismiss this 'computer threat.' The boy cried wolf too many times. 

Yet, there is a reason to take this computer revolution seriously now. Very seriously. To understand why, we need to first learn how computers and the brain work. 

The 'intelligence' of a computer is measured by floating operations per second (FLOPS). Each FLOP refers to one calculation per second. For example, 6 x 3 = 18 is one calculation. A computer with 10 FLOPS, could do 6 x 3 ten times in one second. 

We are now finding out that the intelligence of the brain is also measured in FLOPS. It has been determined that the brain has about 10^16 FLOPS. That's a one with 16 zeroes after it, which is an incredible amount. By comparison, the fastest supercomputer in 2011 had the same amount of intelligence: 10^16 FLOPS. 

So the question becomes: When will your average $1000 computer become as smart as a human? When will it reach 10^16? If you look at trends that have been consistent for 50 years, the answer is: 2023. 

Image Credit: Switchboard. Original source: Kurzweilai.net

Today, a computer for $1000 will give you 2 x 10^12 FLOPS. In other words, it is about 5000 times dumber than you. Yet in the next 10 years, this 5000x gap is predicted to close to zero. This may sound ridiculous, but this type of rapid pace has happened in the past 50 years. Computers follow exponential growth. In other words, they become 10 times smarter every 3 years. 

How are we sure that the brain is ~10^16 FLOPS? The brain is composed of about 85 billion neurons. We'll round up to 100 billion to simplify. Each neuron acts a single calculator and can calculate at about 10,000 times a second. If you take 100 billion times 10,000, you get 10^15 FLOPS. In other words, we are rounding by a factor of 10 up

How can you use this information? We all want to stay employed. The question then becomes: what jobs can robots not take over? Unfortunately, that answer is close to zero. Instead, a better question to ask is: what jobs will robots take over last?

The answer to this is simple: the jobs that require the most amount of intelligence. Jobs that repeat the same task require very little creativity and intelligence. You are given one instruction, and there is no idea generation needed. That is why manufacturing jobs are being threatened by robots like baxter.

However, some jobs require constant idea generation. Thinking of a new Super Bowl commercial, writing the next Hunger Games, or settling a court case all require non-stop idea creation. 

I'll conclude with an opinion. Below are two lists describing what I predict to be examples of "safe" or "in danger" jobs for the next 10 years:

Safe Jobs:
  1. Actors & Fiction Writers -- People want to see real emotions, not 0's and 1's behind them.
  2. Politician -- People want to see a face attached to what they are fighting for. 
  3. Nurse -- When it comes to take care of people, people want emotional beings to care for them.
  4. Lawyer -- When millions of $'s are at stake, people don't settle for less.
  5. Software Engineer -- These are the guys making the computers smart in the first place. 
  6. CEO -- Companies need human oversight and a public figure. 
  7. Police Officer -- To discern between good and bad requires a high amount of intelligence.

In Danger Jobs:
  1. Accountant -- Number crunching is what computers do best
  2. Fast Food Worker -- Machines can cook burgers and much much more. 
  3. Grocery Store Clerk -- Checkout and shelf stocking could be easily be replaced by computers
  4. Truck Driver -- The Google Car is one of many which clearly demonstrates computer power.
  5. News Writer -- This may surprise you, but several software programs scour the web and write their own pieces. 
  6. Construction Worker -- The area of 3D printing has allowed computers to make structures never seen before at a fraction of the cost. 
  7. Teacher -- This is a highly controversial topic. The idea is that the internet has allowed one teacher to teach the masses through free online programs like Khan Academy and Coursera.
© Nicholas Shah and Learning More Than Living, 2013. Please note that an image was used from kurzweilai.net and that full credit goes to them for that image. Unauthorized use and/or duplication of this material without express and written permission from this blog’s author and/or owner is strictly prohibited. Excerpts and links may be used, provided that full and clear credit is given to Nicholas Shah and Learning More Than Living with appropriate and specific direction to the original content. Thank You. 




2 comments:

  1. 10^16 things to think about...(divided by 2*10^15)

    1. Plenty of people live without a smartphone or fast laptop.

    2. FLOPS are not the best way to describe human intelligence. From the WIKIPEDIA ARTICLE you linked (note how it's not from a peer reviewed source):

    Humans are even worse floating-point processors. If it takes a person a quarter of an hour to carry out a pencil-and-paper long division problem with 10 significant digits, that person would be calculating in the milliFLOPS range. Bear in mind, however, that a purely mathematical test may not truly measure a human's FLOPS, as a human is also processing smells, sounds, touch, sight and motor coordination. This takes an average human's FLOPS up to an estimated 10 quadrillion FLOPS (roughly 10 PFLOPS). [1]

    Based on some rough math using OpenWorm as a reference point, modeling the human brain will take in excess of 63.8 Exaflops (10^18). This could be wildly under-estimated as I may not be accounting for the 100 Trillion Synapses[2]. If we are to believe the microtubular computing theory, then this figure could rise to 10^24 [3]. 92.25.228.128 (talk) 15:31, 4 May 2012 (UTC)
    Also note how wide the number ranges are (up to 10^24)

    3. Hunger games is a completed series not requiring any more ideas. I know you didn't know that, but now you do.

    4. You have seen first hand how some people will never be able to use a computer checkout without human assistance.

    5.State policy is always going to impede these predictions whether it is not having laws allowing automated cars or maintaining education regulation.

    ReplyDelete
  2. Some points that aren't FLOPS...

    1. The people who don’t have a smartphone or laptop probably don’t read this blog so the article isn’t addressing them.

    2. It may be difficult to put a number on human intelligence, but even if those estimates are low, computers will still eventually catch up.

    3. It still needed ideas to be generated, and the third movie script is still being edited. You may not have known this, but it was to make up a general example for idea generation.

    4. That doesn’t mean some jobs won’t be replaced with humans trained in interacting with their AI counterparts.

    5. It is not explicitly stated the article was only talking about the USA.

    -ARGotI

    ReplyDelete