What Is It & How Does It Work? [AI/Machine Learning]
We’ve all heard the terms machine learning and AI. They have been in the media for both the right or wrong reasons. Large tech companies have started developing more and more products with Machine Learning in mind. We, at The Whole Caboodle, wanted to lay out what we think the next step in the digital world will be and how it may affect us all.
AI/Machine Learning What Is It?
Artificial Intelligence (AI), is pretty much what it says on the packet – machine-generated intelligence. AI is a growing technology which utilises masses of data, collated from the internet as a whole, in a systematic and automatic way. Acting as the umbrella term for several other aspects of advancing tech, AI encompasses Machine Learning, Deep Learning, Data Mining and Automation…to name a few!
If anybody has seen ‘The Great Hack’ on Netflix, you will already have an insight into the power of AI and *hopefully* realise the importance of the data we put out there as individuals.
Brief History of AI
Back in the 1950s, the founding fathers of the AI field Minsky and McCarthy, described artificial intelligence as “any task performed by a program or machine that, if a human carried out the same task, we would say the human had to apply intelligence to accomplish the task”. Back then, the implementation of this was nowhere near as advanced as it is today and was often used for generating algorithms.
As you can see, AI has been around for much longer than most people think, with current AI systems available over 50 years in the making!
Apple first presented and negotiated the technology of AI in 1991, first launched at TED 1992. Then, after 17 more years of R&D, ‘The Great AI discovery’ was made by three North American scientists. This discovery was known as ‘Deep Learning’ and without this, AI would not be what it is today…
Deep Learning Overview
The core of AI development referred to as ‘deep learning’ is not to be overlooked for its importance, with it touching most modern technologies. This very accurate method of absorbing mass amounts of data works in a very similar way to the human brain, putting 2 and 2 together to make 4, based on patterns and experience. Eliminating human error and processing time, deep learning involves establishing basic parameters around data, using guide tags, so the system understands what the data is and, just like the human brain, learns on its own by connecting the dots and problem solving as it goes.
Just like when the human brain sees a cat. The brain doesn’t think ‘ooh 4 legs and a tail, what can that be?’, you just inherently know, this looks like a cat – albeit a strange one!
Deep learning has superhuman accuracy and self-teaching capabilities. So much so that if we show it masses of sensory data about driving, roads and motorways, it can teach itself to drive as well as a human on those very roads – it might even be better behind the wheel than some folk out there! The discovery of Deep Learning is a subsidiary of Machine Learning, which itself is a subsidiary of AI as a whole.
There are 3 main categories for Machine Learning… Let us break it down:
This process is where humans train an AI system. They provide a large data set that is labelled. This labelling process highlights elements of interest for the system to learn. For example, by providing a machine learning system with 100’s of images of sunflowers, it will know from elements of the image that the object is a sunflower. Now if you provide it with a new set of images of sunflowers (ones it’s never seen before), the machine will be able to identify the sunflower, since it has learned what it looks like from the previous images.
This is where algorithms try to identify patterns within data. It will look for similarities in the data to categorise it. This type of machine learning just groups similarities together, as we see in Google News. You’ll see stories grouping together that have similar topics.
The reinforcement system will look to maximise its reward based on the input data. It does this by going through a process of trial and error until it reaches the best possible outcome. A relatable example would be giving a dog a treat once it has performed a trick. It might take a bit of trial and error until it understands the best action to get the treat but once it understands it will do the trick on the first try next time.
Other Types of AI Explained
As briefly touched on, there are different components to AI applicable for different markets and products.
- Reactive Machines AI – The Main type of AI as we know it. Powered by deep learning algorithms, Reactive AI spots patterns within big data to tailor content and improve the personal experience. Exactly what our Automotive Intelligent CMS & Website is achieving.
- Limited Memory AI – Mainly used for self-driving vehicles.
- Theory of Mind AI – Extremely advanced tech rivalling human intelligence. Remember the cat? This is where the AI pieces together what something is, without being told what it is. This AI is what powers facial recognition technology such as FaceID, Face Filters and advancements in VR.
- Weak AI / Artificial Narrow Intelligence (ANI) – By far the most common AI technology, used by most of us daily. This type of AI powers smartphone features such as Alexa, Siri and Cortana. Simply referred to as ‘Weak AI’ due to its hierarchy within AI. It is just not as strong as we need it to be.
- Strong AI/ Artificial General Intelligence (AGI) – Robots/Live in Doctors/Areas of self-thinking and interpersonal care. This could be implemented within Alexa to combat NHS strain…
- Artificial Superhuman Intelligence (ASI) – Think ‘Humans’ (The Channel 4 series where ‘Synths’ act as housekeepers). This type of AI has the ability to achieve everything that a human can do and more. Self-learning from your behaviour, ASI powered tech can make you feel like it’s a member of the family.
- Self-aware AI – The future of AI development will be even more advanced than the human brain. This divides people from excited to fearful, as we can only wonder how far those self-teaching capabilities will go…well, we have theories but that’s for another time.
“In a way, AI is both closer and further off than we imagine. AI is closer to being able to do more powerful things than most people expect — driving cars, curing diseases, discovering planets, understanding media. Those will each have a great impact on the world, but we’re still figuring out what real intelligence is.”
– Mark Zuckerberg
Drawbacks & Concerns
Naturally, with any new technology, there can be teething issues… Remember Tay (Bot), the Microsoft AI creation with its own Twitter account? Long story short, in the space of 24 hours Tay went from being friendly and optimistic, too – I quote – a “racist asshole” due to the immersion of negative ideologies and bigotry on today’s internet.
Ethics Into AI
Who decides what is good and bad? True and false? What if data was tampered with? We are relying on humans to provide ethical data sources without a bias.
There are several ethical issues with advancements in A.I and Machine Learning. Leading facilities around the world, including The Ethical Center for AI are working on ways to mitigate these, however we feel we should all be aware.
Unemployment – What happens when machines take over manual predictable jobs. For example, when Elon Musk launches his self-driving trucks. What are all the truckers going to do for work?
Inequality – How will new-tech-wealth be distributed? The company owners of AI solutions will take home more and more profits, as they will need less human staff selling time for money. Maybe one solution is that humans own their own AI machine and send it to work on their behalf, giving humanity more time to enjoy the world!
Humanity – How will our behaviour be affected by machines? There is a test called the Turing Challenge – A human talks to an unknown entity and then guesses whether they have just been chatting to a human or a machine. Customer service roles will be carried out by machines and we may never know the difference.
Artificial Stupidity – How do we safeguard against these errors a machine may make that a human would see. We are teaching a machine on how to detect patterns. We can’t provide a machine with every possible example of what it may encounter. This can lead to machine error.
“AI can optimise, not create” – Kai-Fu Lee, Pioneer of A.I for Apple
- Racist Robots – How not to add a humanity bias to AI? Let’s be honest, humans have a bias whether they know it or not, they are the people feeding data to these machines. There will need to be an evolved way to filter out negative and hateful views, which will most likely come from advancements in Deep Learning or Data Mining.
- Security – How to keep AI safe and not to be used against humanity? There is always the risk with such powerful technology that it can fall into the wrong hands and be used for evil over good. How do we stop this?
- Evil Genius – How do we protect humanity from the machine’s efficiency? Example, we task AI to eradicate cancer. The machine goes to work and produces a formula to wipe out cancer from humanity. This formula is efficient, it will eradicate cancer by killing everyone on earth. Obviously, this solves the issue AI was presented with but it’s not how a human would solve this issue.
- Singularity – How do we stay in control? Humans aren’t at the top of the food chain because we are stronger and possess sharp teeth and claws. It’s because we are smarter and can use our intelligence to develop tools to advance. As AI develops it will become more advanced. How do we stop it going beyond humanity?
- Robot Rights – As AI progresses it may move more towards consciousness, where it can feel, or at least mimic feeling and emotion.
- Reinforcing ‘social ideals’ and beauty stereotypes – AI supermodels like Shudu (@shudu.gram) could be reinforcing unrealistic beauty standards with their unrealistic beauty, bodies and the way they interact with real people on social platforms. Also, the fact that tech-giants have insisted to show diversity by making all Social-AI’s black, are tech-giants resurrecting and amplifying aspects of ‘otherness’, which is one of the fundamental basics of racism? Could this, combined with the emergence of size-zero supermodels be doing more harm than good in enhancing the prejudices real humans are working hard to dispel?
Naturally, AI progresses, so do the questions that come with it. How do we integrate it into humanity, without been overrun by robots and technology? Does it make you think of a world like Terminator? We hope not.
How Will AI Change The World As We Know It?
We don’t want to examine this in too much depth; however, here are the two main predictions based on research from market leaders…
Point of View & Censorship
AI can also be used to censor adult content, but in doing this are we limiting humans freedom of speech? In 2016 the Mozilla Foundation commissioned Dutch design studio Moniker to build an AI penis doodle detector, in response to Google’s first open-source AI, ‘Quick, Draw! created as a bit of silly fun with a wider point to prove how much US tech giants control so much of what we see online. Should we be worried that the big corporations get to set the moral standards? The AI created by Moniker and Mozilla is proof of non-consensual censorship…on the doodling of a penis the AI shows a message saying “we assume this was a mistake” and erases it, warning users: “Don’t take individual expression too far!”
(Of course, this is not a full negative – the removal of things like penis drawings may result in a slightly less-filthy world wide web, at least in the eyes of minors and those vulnerable to adult content…we guess?)
Over the next 15 years, repetitive, routine and optimising jobs will likely be replaced by AI. These jobs include telesales, dishwashers, customer support, security guards, truck drivers, haematologists, radiologists, reporters, researchers… However, the current growth of the creative industries is 20% PER YEAR (so we’re good?), as these jobs are the ones which cannot be replaced due to the subjectivity of each role. Complexed and Creative roles, such as CEO’s, Economists, Scientists and Artists are also safe.
We already see AI is finding its way into the workforce. You may have seen those haunting videos of Amazon’s fulfilment warehouses? In such warehouses, you would normally see 100’s or 1000’s of people packing boxes and shipping them out. Now, as you will see below there is a very small number of people involved in the picking and moving/shipping products
Machines Taking over the World?
There have been a few worrying instances of AI going beyond human capabilities. There was the Facebook AI story that broke media outlets, reporting 2 AI bots were tasked to negotiate with each other. Ok, no biggie? Reports claim the tech autonomously found an efficient way of communicating with each other that wasn’t a natural language known by humans or anything the scientists could decipher. So they shut it down. As we know, the media always embellishes the truth.
The official story from Facebook on this story was “While the idea of AI agents inventing their own language may sound alarming/unexpected to people outside the field, it is a well-established sub-field of AI, with publications dating back decades. Simply put, agents in environments attempting to solve a task will often find unintuitive ways to maximize reward. Analyzing the reward function and changing the parameters of an experiment is NOT the same as “unplugging” or “shutting down AI.” If that were the case, every AI researcher has been “shutting down AI” every time they kill a job on a machine.” … However, the sceptics in us may wonder if, like dogs, the machines know how to ‘play dead’.
I think we need to look at both sides of the story, Facebook wanting to play the situation down and the media wanting to earn clickbait. You make your own judgment.
- Customer Service – Improvements in tailored content and the customer experience, will help retailers pinpoint EXACTLY what you want, when you want it – think Clubcard 2.0.
- Improvements in Healthcare – Introducing personalised service to answer simple questions, check your medical records and provide company to the elderly can reduce standard NHS strain easily. Going one step further, utilising deep learning could even lead us to the cures for many illnesses, including cancer (it won’t be the cure we mentioned above, don’t panic!).
- Health & Society – Extensive research says that we as humans, only need to work a 3 day work week for fulfilment. Blockchains, which are linked to the development of AI, can create wealth from no work, meaning a possible increase in ‘pleasure projects’, which in turn could create a friendlier, happier society with health benefits accordingly.
- The democratisation of tech – The democratisation of tech means a more equal marketplace for business – Start-ups building on open-source AI, new ideas being produced and everyone has a chance to learn online.
- Personal Assistant – Alexa & Hello Google helping you with daily tasks. Even Shazam telling you which cool song is on the radio. HeySiri, the list can go on…
- Tailored Content – Tailored online shopping experience such as Amazon. You know what we mean, we’ve all added something to our wishlist then received emails to encourage us to buy with a discount code or the promise of free delivery. This is making its way into the automotive industry, to provide accurate offers on the models you are actually interested in.
- Self-Driving Cars – Autonomous self-driving vehicles are a result of AI. Companies such as Tesla brought them to the market, with AI now being implemented into a range of cars, from low to high budget.
The Whole Caboodle is bringing forms of AI to our website development. Currently, we have built a machine learning CMS that serves tailored content, in a useful way, to an online customer. It doesn’t just stop at tailored website content, we also offer tailored email and SMS communication for our clients and customers. All of this is executed by our newly developed CMS platform, so you can target your potential customers in a way which is personal, specific and best for them.
Make sure to take a look at our Automotive Intelligent CMS section of The Whole Caboodle website. You’ll learn a little more, we also offer a free demo. Just get in touch!