2. Artificial Intelligence
This is the best on the market: Pepper the Robot
Pepper the Robot is programmed to recognise human emotions through facial expressions. It is used in shops and offices to take messages, ‘chat’ with people and sound notifications.
Using AI
A recent global consumer survey revealed that only 34% of consumers think they use an AI- enabled device or service, while 84% actually use AI technology.
Some of the biggest players in the AI industry are the present giant technology companies: Google, Facebook, Microsoft, Baidu, Alibaba, Amazon, Apple, Tesla, IBM and Deep Mind.
But what is AI exactly?
AI- Based in Questions
Question 1 – Can Computers ‘think’?
Question 2 – Can Computers ‘learn’?
Question 3 – Will computers ever match the creative and cognitive abilities of the human mind?
Artificial
Something designed, created, programmed, made by humans
Intelligence
This is trickier to define.
Diverse
AI covers a large area of computer science research, which has grown so large that it encompasses many disciplines making it difficult to limit exactly what it is or isn’t.
For example, Cognitive Computing was once considered part of AI as the goals were so intertwined, but CC has since forked to become its own science.
Definition
“the capability of a machine to imitate intelligent human behaviour.”
Response
It’s a way to program machines or computers to carry out tasks or respond to queries with human intelligence. By taking thousands of data points and setting rules (an algorithm) for the problem-solving process modelled on human neural networks, AI can provide human-like responses.
Real time insights
When IBM’s Deep Blue beat Grandmaster Garry Kasparov in 1997: do you think that this victory meant that Deep Blue was intelligent? Deep Blue had been programmed (by a human) to calculate 200 million possible chess moves a second and had the memory and processing power and speed to calculate billions of move permutations.
But it only had this one task- in that respect it was an instance of what we now call weak (or narrow) AI. Weak AI is one of the key definitions in data driven learning and analysis. Others are:
Key Terms
Strong AI
Machine intelligence that follows the same patterns as human learning: end result being the possession of well-rounded human intelligence
Weak AI
Weak AI tends to be machine learning focused on doing one kind of task. Weak AI might also simply be an intelligent algorithm, which is a set of rules a computer follows to solve a problem
Machine Learning / Deep Learning
A type of AI where a computer can automatically learn and improve from experience without being programmed. Machine learning is really a series of algorithms that give the computer the ability to learn. An algorithm looks at the data and then makes predictions and decisions based on that information
Algorithm
A set of steps that are followed to solve a mathematical problem or to complete a computer process.
Key Terms
Speech Recognition – The ability of a computer to identify human speech and respond to it.
Natural Language Processing- The ability of a computer to understand human spoken and written language.
Internet of Things – The interconnection via the internet of computing devices embedded in everyday objects, enabling them to send and receive data.
The AI Effect
Occurs when something that is AI becomes such a standard part of our experience that we no longer think of it as AI. For example, speech recognition, GPS maps were once considered an essential part of AI. Today, it seems rather normal.
Evolution of AI
1997
Super Computing
IBM’s Deep Blue v Garry Kasparov:
2011
Super Computing
IBM’s WATSON
wins Jeopardy!
2015
Machine Learning
AlphaGo beat Fan Hui (1981–), the European Go champion.
AlphaGo is an AI designed by DeepMind, a company that is now part of Google. Go is an ancient strategy game invented in China more than 2,500 years ago! The rules are simple. Two players take turns laying down black and white stones on a board. If the stone of one colour is surrounded by the other colour, the stone is taken prisoner.
The player who captures the most prisoners and territory on the board wins. Though it sounds simple, Go is much more complicated than chess. In chess, there are 20 possible opening moves. On a Go board, the first player has 361 possible moves!
2017
Deep Learning
The newest version of the Go-playing AI, called AlphaGo Zero, learned to play the game just by playing against itself!
EXAMPLE – AI IN ACTION – IBM WATSON
Watson Imaging Clinical Review improves the path from diagnosis to documentation, eliminating data leaks caused by incomplete or incorrect documentation. This innovative cognitive AI data review tool supports accurate and timely clinical and administrative decision-making by:
- Reading structured and unstructured data
- Understanding data to extract meaningful information
- Comparing clinical reports with the EMR problem list and recorded diagnosis
- Empowering users to input the correct information back into the EMR reports
AI v Machine Learning
- AI despite all the rhetoric really boils down to be a computer program or set of algorithms that does something seemingly clever. It can be a simple knowledge based agent with a set of rules.
- The consensus and general belief is that Machine learning is a subset of AI. Furthermore, the science builds algorithms that allow machines to learn to perform tasks from data that they process or obtain themselves instead of being explicitly programmed.
- Hence the goals of machine learning is to reduce the amount of predetermined knowledge that is imparted to an agent and to let the agent learn about its environment itself through the continual assessment of the data, the precepts, it receives through its own sensors.
- In the last decade, machine learning has produced a deluge of applied AI applications an extremely limited scope of intelligence – such as software robots, which manifest themselves as chat bots, web bots, interactive voice recognition (IVR) systems, and automated software that perform the high-volume repeat tasks like payroll, accounting, finance, order management, and HR in business and loan, claims and mortgage approvals in commerce.
- Many of us also of us have Siri, Alexa, Google Assistant or similar on our phones or in our These devices store the data we provide them with, analyse it through algorithm-based processes and apply machine learning and simple pattern matching to predict our behaviour: movies, music, we might like etc.
- These devices interact with other ‘smart’ devices through IOT sensors: lights, household appliances, watches, cars etc.
EXAMPLE – ENTERPRISE USING MACHINE LEARNING
SonarHome, Poland https://www.facebook.com/sonarhomepl/
SonarHome is a start-up that works in the iBuying model (instant buying), which allows for quick and convenient sale of apartments. This business is based on the platform, which, thanks to machine learning and Sonar Home analytics, enables quick property valuation. Data about localization, size and legal status are confronted with data from popular Polish real estate services lie OLX or Otodom. After getting the value, the SonaHome representative checks the property and negotiates the final price. Then SonarHome buys the property and prepares it for sale. It charges 6 to 10 percent commission for the service of accelerating the sales process.
Source: Sonar Home. Technologiczny klucz do mieszkań, Forbes , Listopad 2019
EXAMPLE – EMERGING AI ENTERPRISE
Voice Lab AI, POLAND https://www.voicelab.ai/
Voice Lab AI is a Polish company dealing with the processing and understanding of speech. The company conducts research and development, creating new algorithms based on artificial intelligence.
One of the main investorS in Voice Lab AI underlines the crucial meaning of collecting data in form of conversation. To develop AI which will be able to effectively process and recognize, huge amount of data is needed. The voice recorded from the radio is not enough.
To teach the AI to recognize the voice, many hours of conversations are required, which differ in transcription, voice and background noise. To understand the scale of the data, it is worth to mention that Google uses 20 thousand recordings to develop its own system.
Source: Czarno na białym. Rozmowa z Jackiem Kawalcem, Forbes. 01/2020
DEEP LEARNING
- Deep learning is a specific machine learning algorithm which automatically learns features, employing a neural network to do so. It is the application of deep artificial neural networks that contain many layers.
- A neural network is called such because at some point in history, computer scientists were trying to model the brain in computer code. The eventual goal is to create an “artificial general intelligence”, a program that can learn anything you or I can learn
- Currently neural networks are very good at performing singular tasks, like classifying images and speech. Unlike the brain, these artificial neural networks have a very strict predefined structure
Source: https://machinelearningmastery.com/what-is-deep-learning/
- The brain is made up of neurons that talk to each other via electrical and chemical signals (hence the term, neural network). We do not differentiate between these 2 types of signals in artificial neural networks, so from now on we will just say “a” signal is being passed from one neuron to another.
- Signals are passed from one neuron to another via what is called an “action potential”. It is a spike in electricity along the cell membrane of a neuron. The interesting thing about action potentials is that either they happen, or they don’t. There is no “in between”. This is called the “all or nothing” principle.
- Thus, we can think of a neuron being “on” or “off”. (i.e. it has an action potential, or it doesn’t)
- What does this remind you of? If you said “digital computers”, then you would be right!
- Binary classification is perfect for the machine learning algorithm of deep learning- May enable the replication of the most profound human experience – ABSTRACT THOUGHT
“Encoded in the large, highly evolved sensory and motor portions of the human brain is a billion years of experience about the nature of the world and how to survive in it.… Abstract thought, though, is a new trick, perhaps less than 100 thousand years old. We have not yet mastered it. It is not all that intrinsically difficult; it just seems so when we do it.”
– Hans Moravec, Mind Children (1988)
IoT: Internet of Things & IIoT: Industrial Internet of Things
IoT has emerged in as a new trend in the past few years where mobile devices, smart transportation, public facilities and home appliances can all be used as data acquisition equipment in IoT: Devices ‘talk’ to one another and relay data – geographical, environmental, logistical.
IoT offers a platform for sensors and devices to seamlessly communicate within a smart network enabled environment, enabling information sharing across platforms: a large number of communication devices are embedded into sensor devices in the real world- and these devices sense and transmit data using embedded communication devices: Bluetooth, Wi-Fi, GSM, RFID.
Over 50 Billion devices expected to be connected in 2030 with the big market drivers being: Internet oriented (cloud), sensors, and data management systems (knowledge)
IoT Big data is different from normal big data collected in terms of characteristics because of the various sensors and objects involved during data collection and complications of hardware automation and embedded systems: subject to the physics of the landscape – need for hardware engineering and material science.
However, implementing IoT could have huge benefits for communication and collaboration, particularly in concepts such as Smart Cities, Smart Retail, Smart Ageing or even the Super Connected home.
What is needed is the next generation of big data technologies that can extract the value from the massive volume of data, in various formats, by enabling high-velocity capture, discovery and analysis. In simplified terms this means that the business opportunities lie in: data sources, data analytics, especially real time analytics and presentation of the results – the Management systems and reporting tools for data.
These big data analytics require all sorts of technologies and tools that can transform a large amount of structured, unstructured and semi-structured data into more understandable data and metadata formats for analysis: algorithms are needed to analyse patterns, trends, correlations etc over a variety of time horizons.