Home > Digitalizacija > Enamul Haque: The Ultimate Modern Guide to Artificial Intelligence

Enamul Haque: The Ultimate Modern Guide to Artificial Intelligence

Understanding Artificial Intelligence

To imagine the future changes, you just need to look at what technologies are emerging and how they will evolve. AI allows business owners to provide a more personalized experience to their customers.

AI is artificially created human-like intelligence. AI is not just about imitating human behavior. In certain areas, AI also includes computers that perform far beyond human intelligence. Some uses are Voice recognition, Image recognition, Automatic Analyzing of Biometric Data. It can also be used to transfer low-recognition data to high-recognition data.

Areas where AI has the most impact in our daily lives are Smartphones, Healthcare (for making diagnoses), Social Networks, Transport, Retail, Navigation, Banking (Fraud prevention, analyzing customer data), Search, Email, Languages (NLP), Advertising (creating relevant content), Service Construction and Entertainment, Virtual Assistants, Surveillance, Commercial Airline Flights.

Applied AI is essentially data analytics and automation.

Some AI companies: AIBrain, Amazon, Anki (robots), Apple, Banjo (search of events), Coca-Cola (Salesforce AI for inventory management), Facebook, Google, IBM (Watson), IcarbonX (personal health), Instagram, Intel, Iris AI (scientist documentation), Next IT (chatbots), Salesforce, Skycode (organizes work queues in industry), SoundHound, Twitter, ViSenze (similar products), X.ai (scheduling meetings), Zebra Medical Vision

AI is generally defined as “a concept or technology for artificially imitating the intellectual work performed by the human brain with a computer”.

AI is a collection of concepts, problems and methods for solving them. Because AI is a discipline, you shouldn’t say “an AI”, just like we don’t say “a biology”.

Machine Learning algorithms force the system to develop independently, studying the available information. Deep learning can strengthen the process of ML. The technology used to do this is called a neural network. Neural network contains numerous layers of compute nodes that work together in an orchestrated manner. We need a lot of data and we need to define what we will need to find.

Data science is a practical discipline that deals with the study of methods for generalized knowledge extraction from data. It is based on methods and theories from many areas of knowledge, including signal processing, mathematics, probabilistic models, machine and statistical learning, programming, data technology, pattern recognition, learning theory, visual analysis, uncertainty modelling, data warehousing.

The digital neuromorphic hardware SpiNNaker has been developed to enable large-scale neural network simulations in real-time and with low power consumption.

AI and IoT are fundamentally changing the picture of the world.

Various programming languages are used to develop AI. AI is currently a program that performs calculation processing using algorithms. The processing of AI:

  • Extract features from data
  • Discover rules from ML and build patterns and models
  • Automatically process new information

When processing vast amounts of data, advanced mathematical statistics and analysis methods are used, so the language of AI is required to have high-speed processing and accuracy.

AI developments in Phyton. The advantage is that you can use Tensor-Flow, Keras, etc. as a ML library.

AI developments in JavaScript. JavaScript is overwhelmingly popular for dynamic web content development. Libraries such as TensorFlow.js, Keras.js and deeplearn.js are available.

AI developments in C++. C++ is a compilation language and is characterized by its high execution speed.

AI developments in R. R language is specialized in statistical analysis.

AI developments in Julia. Julia is a technology calculation language that has been attracting a tension with the third AI boom.

Java Virtual Machine. Java is a language of JVM and it converts and executes programs according to the environment without depending on a specific model or OS.

Haskell. Haskell is the language used for financial investment-related AI because of its bug-free accuracy and security.

GN Octave. There is also a language called GNU Octave for numerical analysis.

Categories of AI:

  • Level-1 Control Program
  • Level-2 Compatible Patterns
  • Level-3 AI That Automatically Learns
  • Level-4 AI That Designs Judgment Criteria

Robotics is a branch of engineering that deals with the construction, operation and application of robots. These robots are programmable devices used to execute a series of commands. Most robot languages are based on languages such as COBOL, BASIC, C language and Fortran.

The difference between AI and robots is whether they can learn based on data.

Humanoid Robots (Humanizing AI). Honda started develop a series of humanoid robots in 1986. It was P2. ASIMO that was developed in 2000 was developed on basis of P2. Boston Dynamics released Atlas. Hanson Robotics developed Sofia in 2015.

In 2017 Sophia was granted Saudi Arabia citizenship.

Speech recognition is another technology that will change the world. It records voice, convert it into digital data and then converts it into text and identifies the speaker.

History of AI

As Pamela McCorduck wrote that AI began with an ancient desire to create a god.

The history of AI research is said to be a repetition of the “boom” and “winter era”.

In late February 2020, Chinese retailers began massively launching order delivery robots to prevent the deadly coronavirus spread. Meiutan Dianping, which initially launched a “contactless delivery” initiative in China has already started using autonomous vehicles to ship groceries in Beijing’s Shunyi area.

Data science and cloud computing

Data science is the science of methods for analyzing data and extracting valuable information and knowledge from them. Data science is intersection between: domain expertise, computer science and mathematics. Domain expertise and computer science intersection is data processing, domain expertise and mathematics intersection is statistical research and mathematics and computer science intersection is machine learning.

Data science is a combination of the areas of mathematics and statistics, programming skills and sector and organizational knowledge.

The five stages of the data science life cycle:

  • Capture: data collection, data input, signal reception, data retrieval.
  • Maintain (support): data storage, data cleansing, data preparation, data processing and data architecture.
  • Process: data mining, clustering/classification, data modelling and data summarization.
  • Analyze: search/confirmatory, predictive analysis, regression, text analysis, qualitative analysis.
  • Communicate (communicating results): data transfer, data visualization, business intelligence, decision making.

Gartner’s 2001 definition of big data is that it is a variety of data that is pouring out at an unprecedented rate. It is referred to as the so-called 3V of Velocity, Volume and Variety.

The development of open-source frameworks like Hadoop (and more recently Spark) has played an essential role in the growth of big data.

Cloud computing is expanding the potential of big data more than ever.

Big data use cases: product development, forecast-based maintenance, customer-experience (CX), fraud and compliance, machine learning and big data, operational efficiency.

Data scientist spend 50-80 % of their time curating and writing data before it is actually used.

The set of approaches and technologies that included tool for massively parallel processing and indefinitely structured data. MapReduce is a model of distributed parallel computing in computer clusters presented by Google. NoSQL is a general term for various non-relational databases and storages and it does not mean anyone specific technology or product. Hadoop is a freely distributed set of utilities, libraries and frameworks for developing and executing distributed programs running on clusters of hundreds and thousands of nodes. R is a programming language for statistical data processing and graphics.

AI and big data are key technologies and factors to advance cloud computing. Many cloud service providers will offer AI capabilities packaged within the cloud services that they vend. Alibaba Cloud’s platform are the Apsara AI Platform, Apsara Big Data Platform and AIoT Platform.

A representative example of AIaaS is the AI speaker. It is difficult to implement high-end HW in a speaker. Therefore, AI speaker manufacturing companies provide voice recognition implemented on cloud servers.

Edge AI is also one implementation of AI. Edge computing includes IoT devices.

Machine Learning (ML)

Machine Learning is positioned as a technology that forms the basis of AI.

An algorithm is a calculation method used when performing calculation on a computer. The basic algorithm in ML is classification. This is a learning method for classifying and predicting information by category. Others are regression (supervised learning), clustering (supervised learning), dimensionality reduction (unsupervised learning), anomaly detection.

To survive ML must have all of the following factors:

  • Data
  • Model
  • Forecasting

The challenge we face today is not collecting or storing the data but analyzing and producing insights that will enable us to give our customers a better service.

In the open-source world, there are two key environments relevant to computational machine development work:

  • Language R – invented in the 1990. R’s development environment is called R-Studio.
  • Language Phyton – Phyton is general programming language used for every purpose, from web development to the control and monitoring of industrial equipment.

The deep ML process consists of two main stages: learning and inference. The learning method should be seen as a method of labelling large amounts of data and defining their respective characteristics. During the inference phase, the system makes certain conclusions and the labels new unexplored data using its previous knowledge.

Basically, AI is umbrella concept, followed by ML and finally Deep Learning, which promises to take AI to the next level. Deep learning is an approach that models the abstract thinking of person.

ANN (artificial neural network) is a mathematical model. ANN consists of artificial neurons, each of which is simplified mode of a biological neuron. The synapse connects them and it has one parameter – the weight coefficient.

Problems solved by Neural Networks:

  • Pattern recognition
  • Classification
  • Decision making and management
  • Clustering
  • Forecasting
  • Approximation
  • Data compression and associative memory

The Advantages of Neural Networks:

  • Solving problems in conditions of uncertainty
  • Resistance to noise in the input data
  • The flexibility of the structure of neural networks
  • High performance
  • Adaptation to environmental changes
  • Fault tolerance of neural networks

In 2017, an Intel subsidiary called Movidius launched the Neural Compute Stick. It has dimensions comparable to a regular flash drive, while inside there is a powerful neural network with deep Machine Learning function.

Convolutional Neural Network (CNN) sounds like an odd mix of biology and mathematics with a touch of computer science.

RNN (recurrent neural networks) are called recurrent because they perform the same task for each element of the sequence, with the output depending on previous computation.

Internet of Things (IoT)

IoT is a mechanism that connects things around us via the Internet. It is also possible to install sensors in the IoT so that it can grasp its own state and environment and collect data through the network to utilize the data. The data is called IoT data.

Three factors are decisive for machine-machine communication:

  • Data integration point (DIP): a server that manages and stores incoming data.
  • Data End Point (DEP): a networked car, a gas/water meter, a container with a chip or a logistic robot.
  • Transmission technology: WLAN, Bluetooth, landline, satellite, radio or RFID.

Difference between AI and IoT is Hardware and Data. AI flows from software to hardware. In IoT processing from hardware to software is the main process.

Linking IoT and AI:

  • Autonomous Driving
  • Manufacturing Industry: production status, inventory management, equipment utilization rate and predictive maintenance.
  • Utilization in the Field of Agriculture
  • Smart Refrigerators
  • Environment, Safety and Economy

The use of AI

AI can improve or expand existing methods in IT product and service management:

  • Service desk
  • Ongoing Monitoring
  • Incident Management
  • Answering Inquiries
  • Problem Management
  • Reactive Tools
  • Proactive Tools
  • Predictive Tools
  • Autonomous Tools

DevOps is a method in which development and operations work closely together by using the same development methods and tools to efficiently build a system. DevOps has three benefits: the first is improved reliability, the second is to improve productivity and the third is flexibility.

The advantages of using AI Ops:

  • Incorporate processing speed and accuracy into operations
  • Automate manual work to increase productivity
  • Speed up problem repair

Typical AI Ops tools: Splunk, OpsRamp, Watson OpenScale.

RPA (Robotic Process Automation) is the use of AI software and ML resources for machine maintenance. This may also include querying, calculating and managing records and transactions. software robotics means the automation of repetitive routine works.

RPA life cycle:

  • Analysis
  • BOTS – Bots development
  • Tests
  • Support and Maintenance

Types of automation: software and industrial. There are five areas where RPA takes different approach than traditional automation: technology, support/software restriction, record and play, reading and customization.

AI has become an essential part of modern weapons.

AI is also used in agriculture. Sensors and UAVs are used to monitor growth. According to Zion Market Research Institute the market for AI in the agriculture will grow to 2,1 billion USD in 2024.

Watson Career Coach is used by IBM employees to get career coaching.

AI is also used for OCR, customer services, inside our homes.

The Future of AI

Some areas of AI impact in future will be: transport, manufacturing, healthcare, education, media and customer service.

Singularity was not originally a term-limited to the field of AI. It is a term used in mathematics and physics and cultural anthropology and humanities as a “cultural singularity”. A singularity in mathematics or physics is a point where specific criteria cannot be applied.

Ray Kurzweil lists of three revolutions – GNR. G is Genetics, N is Nanotechnology and R is Robotics.

5G networks will be fast and able to support many more devices simultaneously than the network that came before. By 2025 there will be 41.6 billion devices connected to IoT.

AI regulations and enforcement will become urgent.

Future possibilities:

  • Overcome cancer, myocardial infarction and stroke
  • Communicate with people from all over the world with just one headphone
  • Freedom from housework – one domestic robot per family
  • Going cashless
  • Child safety protected by the sensor network
  • Cars don’t bump into each other

Jobs that could possibly be replaced by AI in the future:

  • Telemarketers
  • Bookkeeping clerks
  • Benefits manager
  • Receptionist
  • Couriers
  • Proofreaders
  • Computer support specialist
  • Market research analyst
  • Advertising sales
  • Retail sales
  • Accountant
  • Security Guard

Jobs created by AI:

  • Specialist in Empathy
  • Algorithm interpretation
  • AI security specialists
  • The resilience specialists

American IT company Cognizant is talking about 15 new occupations:

  • Data detective
  • Genome portfolio director
  • A partner for a walk/conversation
  • Ethical Procurement officer
  • Chief credit officer
  • Cybercity analyst
  • Human-machine collaboration manager
  • AI business development manager
  • PEU (personal equipment utilization) IT facilitator
  • Edge computing expert
  • Fitness commitment counsellor
  • Digital tailor
  • AI-supported medical technician
  • Financial soundness coach
  • Quantum ML analyst

The fact is that there will be much closer cooperation with intelligence machines in the future. Only those who also participate in this digital change will be a wheel in the gears of working society in the future.

It has been said that human intelligence, or smarter AI, is the last invention that humans need to make.

You may also like
Mustafa Suleyman, Michael Bhaskar: The Coming Wave; AI, Power and Our Future
How will leading people look in the AI world
Thomas M. Siebel: Digital Transformation; Survive and Thrive in and Era of Mass Extinction
Bill Schmarzo: The Economics of Data, Analytics and Digital transformation

Leave a Reply