AI, Machine Learning, NLP, and Deep Learning?
Are these ingredients the future of human success?
Autonomous or driverless vehicles are a hot topic on the AI scene right now. Google, Volvo, Tesla, Uber… these are just some of the big names in the race to prove that driverless or autonomous vehicles are better and maybe even safer than human-driven vehicles.
I was at a family event recently and two guests were chatting about the Artificial Intelligence (AI) component of driverless or autonomous vehicles and more specifically, how these vehicles are currently unable to detect human movement at high speed. One cited the example of a child stepping into the road whilst a vehicle was approaching at high speed. Some debate ensued about the width of lanes in the road (surrounding the vehicle) and the impact they have on the judgement of the driverless/autonomous vehicles. The discussion evolved to include the concept of ‘human thought’ and ‘human mental processing’, relative to that of the machines/AI, and by parallel, if the machines/AI would make the same ‘judgement calls’ as people do and if machines/AI were not currently capable of doing so, then when.
One said: “AI will never be capable of ‘thinking’ as we humans do and in fact these vehicles are not ‘Artificial intelligence’ at all but mere algorithms, computer programming, and this is why the vehicles cannot adapt, they require programming”.
The other guest suggested AI could learn and adapt from one satiation to the next thereby teaching itself how to better handle the situation, over time. The other corrected them stating that it is not possible. “AI is in no way comparable to the human mind, it cannot learn, AI is simply algorithmic by human design”, he said.
I challenged both asking if they had seen the film Ex Machina (2015)? This film depicts the very essence of this discussion whereby, a young programmer is selected to participate in a ground-breaking experiment in synthetic intelligence by evaluating the human qualities of a breath-taking humanoid A.I. and the film opens the imagination of its viewer of the power behind possibilities of machine learning.(Ex Machina — film review courtesy IMDb.)
There was some confusion between these two people due to a lack of deep understanding on the topic.
AI, Bots and Chatbots and in turn through them, the Internet of Things (IoT), are slowly beginning to ingrain themselves in our daily lives, taking full responsibility of our lives (as is the case of driverless vehicles) and in some cases intimately listening or overseeing our every move for example: in the home, with Amazon’s Echo (including Amazon Alexa the home automation PA) or in your phone, with Google Assistant or Siri…(and the list goes on).
Wikipedia is always a good gauge of communities opinion into the meaning of things (whilst it tries to be the truth, it is not always the factual truth). So, and out of curiosity I Googled driverless vehicles. Wikipedia came up as it usually does but it was incredible to see that the latest update to Wikipedia’s definition of Autonomous Vehicles (aka Driverless Vehicles) was 28 Feb 2017. This implies that the definitions, as society are defining them (in this field) are literally evolving as I type.
It was then I realised the need to expand on some fundamental concepts, the key ingredients of AI and the core differences between them:
AI, Machine Learning, NLP and Deep Learning
Artificial Intelligence
According to Merriam-Webster dictionary, artificial intelligence is defined as :
1: a branch of computer science dealing with the simulation of intelligent behavior in computers
2: the capability of a machine to imitate intelligent human behavior
Oxford Dictionary define AI as:
“The theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages.”
To my mind, artificial intelligence is under continual exploration so the ‘theory’ is no more. As the simplicity of definition 2 suggests by Merriam-Webster, AI is no longer just a theory. Machines are now capable of and imitating intelligent human behaviour. Naturally and imitating their creators, current AI are built to evolve. Ultimately, man is looking to empower computers to approach reasoning (to problem solving) and conquer everything in life, just as we humans do.
Version upon version, AI continues to develop all the time and the process is iterative and built upon improvement, learning from mistakes of the past, one model at a time. Learn, build, learn, grow and so the cycle continues, just like a baby becomes a child who becomes an adult.
To make the concept that is AI as simple as possible we can examine the basic components that go into making an AI being. In other words, how does AI go from a bunch of technology things to actually doing stuff for us?
I play around with and build bots and chatbots. If you’ve missed my previous articles, you can catch up here. In context of Bots and Chatbots and for the purpose of this article, I’ll use the basic diagram below to conceptually explain the components.
Form left to right, the diagram depicts a Facebook messenger bot or just as easily an independent purpose build bot placed on the homepage of a website and with the purpose of connecting with customers in a customer service capacity providing and AI-like customer service experience or response to customer enquiries. In the centre is the NLP and Deep learning are and to the right and below the connectors and storage, allowing the bot to access a store of all the info is will need and in order use it to learn and grow.
Machine Learning
Where does the Bots (or AI) intelligence (thought processing) come from, how does a machine learn? Man programs this into the architecture in the form of a vast array of algorithms to iteratively automate the process of learning. The computer uses the algorithms to assess actions taken and decision made and their impact over time and in turn, at the same time, the machine learns.
Wikipedia has this to say about Machine Learning: “Machine learning is a subfield of computer science that evolved from the study of pattern recognition and computational learning theory in artificial intelligence. In 1959, Arthur Samuel defined machine learning as a “Field of study that gives computers the ability to learn without being explicitly programmed”.
So, if machines are capable of learning and we are programming them to ‘think’ and respond, this begs the question: how can they learn to interpret the natural tongue and vast array of human beings different languages and the slang that we humans speak, across the globe? In addition, how can they understand, make conversation and put the ‘human’ back into the conversation and in human language (not algorithmic slur), just like humans do? This is where NLP comes in.
NLP — Natural Language Processing
NLP — Natural Language Processing is the component that bridges the gap between human talk and computer programmed understanding (regardless of which human is talking, what language is spoken and the way in which, perhaps grammatically, they are talking). The NLP component allows the computer to interpret the vast and complicated human language, understand what’s being said, process it all, reflect what is being required of it and effectively ‘talk back’, equally like humans do. Through the process, the machine has to understand all the lingo being used and develop or adapt with the ability to reply, not as a computer but rather a human-like computer. NLPs are advancing at exceptional rates and it is this human-like and natural, behavioural part of the experience, the NLP handles.
Deep Learning
Deep learning is a deeper level and subset of machine learning whereby a machine uses mass amounts of data and highly complex algorithms to ‘learn’ and to simulate human-like decision-making. The learning area within an AI is under massive development and growth currently coining the term, deep learning.
Deep learning is the deeper part of the AIs brain and the conduit to the NLP allowing the machine to learn from everything and improve upon itself for next time, just as humans would.
For example, if you grab your common iPhone and ask Siri if she’s happy, she answers: “I’m happy, I hope you are too..”. If you ask her why or what makes her happy, her answers are nonsense — she is not capable of an adequate human-like response. I believe that Siri is not built with Deep Learning capability yet but I do believe the iPhone and AI part of the iPhone of the future will harness all of our voice activated engagements (through the life of the phone) and can and will assess and analyse all conversations and behaviours (as exhibited through our actions and words), learn our mood and feeling (as expressed through our speech) and in turn could respond why it was happy and, what could make it happy (just like a child learns happiness from listening to laughter). In other words, the computer could use the algorithms we provide mixed with our responses to learn ‘how’ a human behaves and responds and mimic doing so, all by itself and just as expected from the prefect AI machine. This is the essence of deep learning.
There are two current well-known products in the market leading the charge on deep learning capabilities: Microsoft’s LUIS and IBM’s Watson (and no doubt many more are being developed as I type).
A good example to demonstrate what NLP and Deep Learning working together represent has been put very well by the Director of Product Strategy, Retail solutions provider at IBM, regards Watson:
“ When people ask how Watson is different than a search engine, I tell them to go on Google and type ‘anything that’s not an elephant.’ What do you get? Tons of pictures of elephants. But Watson knows those subtle differences. It understands that when feet and noses run, those are very different things”.
This principal has differentiated LUIS and IBM’s Watson based on one major factor or difference: Cognitive. Cognitive-based systems build knowledge and learn, understand natural language, and reason and interact more naturally with human beings than traditional programmable systems. The term “reasoning” refers to how cognitive systems demonstrate insights that are very similar to those of humans. (Ref: IBM Institute for business value — Your cognitive future — part I).
These areas of AI and the various components are all just the beginning. Man has been studying behaviour, engagement and now mechanics of the brain, since the dawn of time and since the start of our very existence. We have used our deep learnings from our research and exploration to develop AI. Imagine how many more ground-breaking components will be exposed and how fast this can and will take shape especially with the assistance of our AI friends, by or side. Imagine a world where, for example, we could forecast the need for a new vessel to hold food and drink. The AI could assist by running through permutations even we have not been capable of thinking of yet and then presenting us with the blueprint of how to produce such an economic vessel…the future will be filled with these augmented experiences and all thanks to the scientists of our time for developing todays AI, machine learning, NLPs and Deep Learning.
I have a great passion for these new technologies but I appreciate most people do not. I hope you’ve enjoyed this article and a deeper insight and understanding into the DNA of AI and in simple terms grasping that using deep machine learning and a natural language processor, a (computer) being can become artificially intelligent.
Article written by Glenn Miller. Please visit my Website and feel free to connect. I love speaking at conferences and business forums, meeting new, innovative and exciting people.
