Today we hear the buzzword of Artificial Intelligence. You can use AI to combine colors, AI to replace the sky in your photo, and AI for your smart homes. It seems they’re helping us a lot, but at the same time, people are scared that AI will take over humans’ tasks, leaving us jobless.
The lingering question becomes whether it is good or bad. How can AI improve the economy and businesses without creating negative effects? And, of course, the real question is: what exactly is AI?
This post contains information you need to know about AI and how you can use it to your advantage. First, we look at the history of computers, then move to how they think and process data and daily applications. The goal is to give you ideas to use AI in designing your apps and user experience.
Historically, computers were born to compute
A basic computer is like a calculator: it computes whether a value is equal to another. Usually, we assign a sequence of computers to do, and they evaluate it through computational methods. Previously, it cost 19 billion dollars for 1 Giga FLOPs, but in 2017 the cost became 3 cents only. Computers become faster, cheaper, and more advanced as time goes by. Read more on how you can stay on top of your competition by continuously improving the product UX with AI technologies.
Algorithms: the secret behind good programming
Google, Facebook, Amazon, online dating websites, and MP3s are examples of how algorithms can be used. The basic idea is to transform information into useful insights using algorithms. For example, with Google PageRank, the program presents a list of website links sorted by relevance to your search. For MP3s, music is compressed from 32MB Compact Disc storage to 3MB songs for easier storage. Read more about how algorithms can advance the UX in our previous article.
Then we predict the future...
With the technologies we have, we can give future forecasts. We can predict a future trend in sales or expenses or any relatable numbers with massive amounts of data from historical data or similar industry companies. On Netflix, for example, there are recommended movies for users to watch. This is based on their previous behaviors and what others are watching. Check out this article about predictive analytics and its role in exceeding customer expectations.
Massive data is the key to data science
Interconnectivity is important for data science. A smart home would have multiple sensors that can be activated with a voice command. With sensors, a smart city can manage security surveillance and improve services, such as public transportation. Tesla produces cars with multiple sensors and manages them with software. Other automotive companies must check the overall system, call back their cars, and re-manufacture them when a defect is found. Meanwhile, Tesla can look at the problem, fix the software issues, and upload it to the cloud so that customers can download the new version. Data science is the key to integrated solutions that benefit customers and producers.
The ability to learn and interpret data is neural networks
Like the human mind, neural networks are algorithms that train computers to understand what data is given to them. For example, when a cat's picture is thrown into a computer's memory, it should be able to identify it as a cat. A simplified version of this case is to throw scribbles of the number ‘3’ and let the computer learn that it is the number ‘3’. The computer would look at the image grid-by-grid and determine what it is based on the darkened grids—more on neural networks.
Machine Learning vs. Deep Learning
Computers are good at repetitive tasks. We can train AI to determine objects through repetitive tasks. One example is to determine what is a ‘good’ flower that farmers can pick. The AI technology can receive many flower images and match them with ‘good’ flower criteria. The program learns from trial and error as it interacts and receives human feedback. This is Machine Learning at the core.
Meanwhile, Deep Learning involves larger input data and more advanced results with higher accuracy. Training AI with deep learning will take longer, but its outcomes can differ. Read more about the differences- Machine Learning vs. Deep Learning.
Chatting with computers is not as easy as it seems
Computers can mainly analyze and produce speech and text. We have seen it in action while calling banks and inserting numbers to pick an assistance category based on our needs. We see it in Chatbots that pop up to help save us time rather than scrolling through the website's pages. Computers are still learning to speak to us through Natural Language Processing.
Internet of Things or the platform for AI
Multiple sensors build our world: motion, temperature, light, wind, etc. Computers assess the data from these sensors to give feedback or results. While it goes unnoticed, our mobile phones have motion sensors where the screen lights up whenever we use it. We also see Home devices activate when there is a change in room temperature. The key is to enhance connectivity to improve efficiency—more on the Internet of Things (IoT).
Edge AI helps build local, responsive algorithms
We can also apply AI algorithms when computers are offline (locally). Edge AI gives companies the advantage of privacy, security, latency, and load balancing. An example would be Edge AI in Tesla cars that can make a split-second decision when the car is about to crash. Another example would be maintenance apps on airplane hangers with low connectivity. Read more about Edge AI.
Designing a better AI with UX Design
Ultimately, we want to provide a good user experience to our customers. The idea is to engage the customers at multiple points: before they know your company, as they show interest in your products and services, as they become a customer, as they become repeat customers, and so on. UX Design in AI helps contextual Artificial Intelligence, considering users' ecosystem. Contact us to see if Artificial Intelligence must be part of your UX strategy.
Editor's note: This post was originally published in December 2020 and has been updated for comprehensiveness.
