IBM Q(uantum)
I've been here before, but I think I was a bit too early....
http://tonyisageek.blogspot.com/2017/05/babies-first-computer-quantum.html
However, I'm pretty sure that this is the place to start to move over into.
If you think about the limitations we have today in relation to Machine Learning and Artificial Intelligence, it's a bit like looking at the limitations of the horse&cart and having the vision to see the Quantum Sports car (okay, not the best evolutionary step, but a vision none the less)
During my day job, I've seen a lot of people working away delivering various Machine Learning services, such as "chat bots". They are all super excited that they are going to really make a difference and help out their fellow human beings to "get things done".
Then they release their baby into the wild and well, it gets a whole lot of abuse and gets condemned for being "dumb", for not being able to do the "simplest of things". One thing that the end users (who I can usually refer to as "the dumb one's") miss the point on is that it is (mostly) still based on rules-based computer software configured by a human being. It is not an Artificial Intelligence, it is not a human, it is not going to magically learn and "just know stuff".
Why not? Processing power. We've got the ability to gather a lot of data. A lot of data! Hard drive space is cheap, speeds of read/write is blindingly fast (with SSD), so gathering the data to acquire knowledge is not the problem.
Writing algorithms to make sense of all that data has a niche placement (thanks to all the geeky Data Scientists who have a weird fascination for mathematics and solving challenging problems) but once you have the formula for those algorithms and potentially overlay multiples you can start to build out models that "make sense" of that data, so that it can be used as a ground truth for "knowledge".
Keeping that quantity of data, sanitised, meaningful and non-garbage is a challenge. That's why attempts at making a "real" A.I. and releasing it into the wild have been publicly ridiculed (you can google just as well as I can). There still needs to be a lot of hand-holding, nudging, nurturing, I suppose it is not too dis-similar to a human parent and child, there is a huge dependency on the parent during the initial years, to help with the ground truth and to set some guidelines for filtering of that data.
When you now want your chat bot to be as knowledgeable as a "real" A.I., to be able to be provided with enough ground truth to make decisions with confidence, to learn from historic decisions (for good or bad) and to not end up becoming the next Skynet, what are you going to do?
There is also another aspect to consider, Machine Learning leads you to Artificial Intelligence. Think about it. Are humans always correct and accurate immediately? no. So, why would you expect a piece of computer software that is built around similar rules as humans to be correct all of the time?
One of the fascinating things with my day-job is having to point out to people that whilst their rules-based chat bot was confident of understanding x amount of utterances, categorising them into Intents and working out what the end-user is trying to achieve, that was "then". ie. that was 100,000 "discussions" ago. The chat bot has "learnt" a lot since you gave it the ground truth to build up from. Some of the initial ground truths have moved and an answer based on today's understanding might be different to the same answer a year ago. It's not all 1's and 0's anymore.
And there is the mind-shift for the next generation of computer engineers. We originally grew up in the I.T. world based on a 1 or 0, a True or False (true/false) binary based logic, that was kind of fed from Science/Chemistry lessons where you would perform your test and analysis and based on the rules the outcome should always remain the same and therefore the truth will be proven.
Well.....it looks like it's time to shake things up a bit now.
As stated above, the world of the 1 and 0 is going to be looked back on like the horse&cart . We're going to be venturing into the world of the Bugatti Veyron in our life time.
How is that going to be possible? Welcome to the world of QUANTUM COMPUTING (not some pipe-dream on the SyFy of Discovery channel, but a real working, functional and useful computer).
This was an article I originally read a year or so ago:
https://www-03.ibm.com/press/us/en/pressrelease/51740.wss
Artificial Intelligence: Making facets of artificial intelligence such as machine learning much more powerful when data sets can be too big such as searching images or video"
I had a play around with the original machine as anyone could:
"The IBM Q Experience is an online platform that gives users in the general public access to a set of IBM's prototype quantum processors via the Cloud, an online internet forum for discussing quantum computing relevant topics, a set of tutorials on how to program the IBM Q devices, and other educational material about quantum computing"
https://www.research.ibm.com/ibm-q/
You could even use the IDE (which reminds me of music scores/sheet music):
https://quantumexperience.ng.bluemix.net/
Now, marry that up with the most recent Apple Inc, news:
https://www.wired.com/story/how-apple-makes-ai-chip-powering-iphones-fancy-tricks
So there we have both the Server-side (IBM Q) powering the sub-conscious brain elements and the Client-side (A.I. chips in smartphones, etc..) powering the conscious elements, we could start to see some joined up evolution going on.....finally.
All very metaphysical and potentially a representation of a view of human existence:
Or, for the more Science Fiction orientated, how about some good old fashioned classic Cylons:
I for one, shall be looking to re-align my skillsets to adopt the new shift in computer hardware and software to take full advantage of what is coming up on the horizon.
There is some quote from some movie that goes along the lines of, "Well, I'm figuring out how to build and make them, so when the time comes I'll be one of the lucky ones who knows how to turn them off". (something like that).
http://tonyisageek.blogspot.com/2017/05/babies-first-computer-quantum.html
However, I'm pretty sure that this is the place to start to move over into.
If you think about the limitations we have today in relation to Machine Learning and Artificial Intelligence, it's a bit like looking at the limitations of the horse&cart and having the vision to see the Quantum Sports car (okay, not the best evolutionary step, but a vision none the less)
During my day job, I've seen a lot of people working away delivering various Machine Learning services, such as "chat bots". They are all super excited that they are going to really make a difference and help out their fellow human beings to "get things done".
Then they release their baby into the wild and well, it gets a whole lot of abuse and gets condemned for being "dumb", for not being able to do the "simplest of things". One thing that the end users (who I can usually refer to as "the dumb one's") miss the point on is that it is (mostly) still based on rules-based computer software configured by a human being. It is not an Artificial Intelligence, it is not a human, it is not going to magically learn and "just know stuff".
Why not? Processing power. We've got the ability to gather a lot of data. A lot of data! Hard drive space is cheap, speeds of read/write is blindingly fast (with SSD), so gathering the data to acquire knowledge is not the problem.
Writing algorithms to make sense of all that data has a niche placement (thanks to all the geeky Data Scientists who have a weird fascination for mathematics and solving challenging problems) but once you have the formula for those algorithms and potentially overlay multiples you can start to build out models that "make sense" of that data, so that it can be used as a ground truth for "knowledge".
Keeping that quantity of data, sanitised, meaningful and non-garbage is a challenge. That's why attempts at making a "real" A.I. and releasing it into the wild have been publicly ridiculed (you can google just as well as I can). There still needs to be a lot of hand-holding, nudging, nurturing, I suppose it is not too dis-similar to a human parent and child, there is a huge dependency on the parent during the initial years, to help with the ground truth and to set some guidelines for filtering of that data.
When you now want your chat bot to be as knowledgeable as a "real" A.I., to be able to be provided with enough ground truth to make decisions with confidence, to learn from historic decisions (for good or bad) and to not end up becoming the next Skynet, what are you going to do?
There is also another aspect to consider, Machine Learning leads you to Artificial Intelligence. Think about it. Are humans always correct and accurate immediately? no. So, why would you expect a piece of computer software that is built around similar rules as humans to be correct all of the time?
One of the fascinating things with my day-job is having to point out to people that whilst their rules-based chat bot was confident of understanding x amount of utterances, categorising them into Intents and working out what the end-user is trying to achieve, that was "then". ie. that was 100,000 "discussions" ago. The chat bot has "learnt" a lot since you gave it the ground truth to build up from. Some of the initial ground truths have moved and an answer based on today's understanding might be different to the same answer a year ago. It's not all 1's and 0's anymore.
And there is the mind-shift for the next generation of computer engineers. We originally grew up in the I.T. world based on a 1 or 0, a True or False (true/false) binary based logic, that was kind of fed from Science/Chemistry lessons where you would perform your test and analysis and based on the rules the outcome should always remain the same and therefore the truth will be proven.
Well.....it looks like it's time to shake things up a bit now.
As stated above, the world of the 1 and 0 is going to be looked back on like the horse&cart . We're going to be venturing into the world of the Bugatti Veyron in our life time.
How is that going to be possible? Welcome to the world of QUANTUM COMPUTING (not some pipe-dream on the SyFy of Discovery channel, but a real working, functional and useful computer).
This was an article I originally read a year or so ago:
https://www-03.ibm.com/press/us/en/pressrelease/51740.wss
"IBM intends to build IBM Q systems to expand the application domain of quantum computing. A key metric will be the power of a quantum computer expressed by the “Quantum Volume”, which includes the number of qubits, quality of quantum operations, qubit connectivity and parallelism. As a first step to increase Quantum Volume, IBM aims at constructing commercial IBM Q systems with ~50 qubits in the next few years to demonstrate capabilities beyond today’s classical systems, and plans to collaborate with key industry partners to develop applications that exploit the quantum speedup of the systems.
IBM Q systems will be designed to tackle problems that are currently seen as too complex and exponential in nature for classical computing systems to handle.
Artificial Intelligence: Making facets of artificial intelligence such as machine learning much more powerful when data sets can be too big such as searching images or video"
I had a play around with the original machine as anyone could:
https://www.research.ibm.com/ibm-q/
You could even use the IDE (which reminds me of music scores/sheet music):
https://quantumexperience.ng.bluemix.net/
Now, marry that up with the most recent Apple Inc, news:
https://www.wired.com/story/how-apple-makes-ai-chip-powering-iphones-fancy-tricks
So there we have both the Server-side (IBM Q) powering the sub-conscious brain elements and the Client-side (A.I. chips in smartphones, etc..) powering the conscious elements, we could start to see some joined up evolution going on.....finally.
All very metaphysical and potentially a representation of a view of human existence:
"As above, so Below"
Or, for the more Science Fiction orientated, how about some good old fashioned classic Cylons:
I for one, shall be looking to re-align my skillsets to adopt the new shift in computer hardware and software to take full advantage of what is coming up on the horizon.
There is some quote from some movie that goes along the lines of, "Well, I'm figuring out how to build and make them, so when the time comes I'll be one of the lucky ones who knows how to turn them off". (something like that).
Comments
Post a Comment