Besides Blockchain, AI must be the most overused term in the industry today. It has been popularized by movies such as the Matrix and Ex Machina, as well as by utopians like Mark Zuckerberg and dystopians like Elon Musk.
But what is artificial intelligence, and how does it differentiate from machine learning and deep learning which are often used interchangeably?
Simply said: Artificial intelligence (AI) is the ability of a computer program or a machine to think like humans do.
AI is using technology to do things which used to require human intelligence, for example problem solving or learning, and will likely shape our future more than any technology that has come before.
Machine learning (ML) is a subfield of AI and gives machines the skills to ‘learn’ from examples without being explicitly programmed to do so. Deep learning (DL) is a specialized ML technique that mimics the behavior of the human brain and enables machines to train themselves to perform tasks.
🤯 AI is already able to defeat the world’s best poker players, predict crime, read your lips, clone any voice, beat human doctors on clinical exams as well as better predict the death risk of their patients.
As should be clear from the above, machines are increasingly getting smarter and are picking up more and more tasks which could previously only be done by humans.
A lot of people therefore have this misconception that artificial intelligence was designed to replace humans and whatever we do in our daily work or at home. However, AI was (and is being) developed for the sole purpose of augmenting our lives and amplifying our skills and capabilities in all that we do. We are entering an age where man and machine will collaborate ever more closely. Many countries have published or announced national AI strategies to capitalize on the technology’s potential and become a global leader.
Image via becominghuman.ai
There are three main algorithms or techniques you should be aware of:
Big data acts as an ingredient. Think of it as when you are making a cake – the data represents the flour and the actual process of baking the cake is represented through machine learning. Artificial intelligence will then be the output, or the cake if you will.
AI enables speed and accuracy levels previously impossible. We are experiencing massive growth in unstructured data from numerous sources from an always-on consumer. The ability to highlight non-obvious relationships in these data sets due to rapid data processing and analysis drastically improves decision making at scale. This will help organizations with future planning as they can learn from past behaviors and act in real-time.
Besides allowing for hidden pattern analysis and better prediction models, automated learning enables hyper-personalization. Use cases go from marketing to a segment of one (the holy grail of every marketeer) to areas such as personalized food and precision medicine.
Increased productivity and operational efficiencies are key reasons behind the widespread adoption of AI. Using advanced modeling and simulation techniques, machine learning can assess automated processes in the organization and make any adjustments in real-time.
Mundane tasks that do not require time to think will be automated first. This could lead to a reduced head count while time gains will result in important cost savings. Improved labor productivity as a result of AI investments will give employees the opportunity to focus on more creative and value-adding work.
Whilst the efficient analysis of big data sets and more ‘objective’ decision-making will be a main advantage of artificial intelligence, there is a long road ahead to solve for the bias problem of algorithms. You may have heard of Microsoft’s racist chatbot Tay which took a turn for the vulgar in less than 24 hours on Twitter, or the case of risk assessment software COMPAS which predicted blacks were twice as likely as whites to be labeled a higher risk of recidivism.
The main problem is that we humans are inherently biased. AI hence reflects and augments all of our society’s assumptions and perceptions – good or bad.
With AI increasingly being used in the hiring process and in fare pricing by companies such as Uber and Lyft, prejudice needs to be tackled head-on if consumer trust is ever to be gained.
Artificial intelligence (AI) and machine learning (ML) have wide applicability, going from your Netflix content recommendations to detecting fraud with PayPal and even allowing drones to fly autonomously.
Let’s look at the role they are playing as the next big industry disruptors.
Banking and insurance are leading the way in the adoption of artificial intelligence. FinTech and InsurTech are disrupting the industry helping reduce administration time and automate back office operations.
A popular use of AI you might be familiar with is the personal digital assistant. Chatbots and voice assistants are being used to provide 24/7 customer service and let customers check their account balances and make money transfers.
Active.ai and Finn.ai are some of the platforms specialized in developing these virtual assistant solutions for financial institutions.
Robo-advisors like Betterment simplify investing and wealth management and make such services more accessible to a wider population, while HSBC is using AI to prevent fraud and money laundering by spotting abnormalities in transactions.
The use of AI in retail is heavily focused on providing a better in-store and online customer experience. Extreme personalization is made possible due to the real-time analysis of big volumes of data.
Besides virtual assistants and voice-based shopping, artificial intelligence is being used by retailers like H&M to predict the next big fashion trends and customize their store offering.
Other AI use cases address bottlenecks in the supply chain such as inventory management, demand forecasting and automatic product replenishment. German retailer Otto is for example partnering with Blue Yonder for price optimization and reducing their out-of-stock rates.
For an in-depth look at how artificial intelligence and other exponential technologies are disrupting the retail industry, go here.
One of the most prominent examples of the use of artificial intelligence can be found in assisted and autonomous driving. Every big car manufacturer is heavily investing in AI to improve driver safety by automating the detection of pedestrians and road signage, and creating the ability to anticipate and course correct when something goes wrong.
Affectiva launched Automotive AI, a platform which can sense driver fatigue, anger or distraction and alert or intervene accordingly. In-car music can be personalized, and lighting and heating can be adapted based on the analysis of facial and vocal expressions of passengers.
Intelligent diagnostics is a game changer in the medical industry. Its main power lies in the ability to analyze vast amounts of information such as medical history and genetic information across systems inside and outside the healthcare organization. AI has already proven to be better than dermatologists at detecting skin cancer and can spot Alzheimer a decade before human doctors. Equipped with AI, your Apple Watch can even predict heart problems and detect diabetes.
Pharmaceutical giant Pfizer is developing a drug discovery platform, while LG Health and Aspire have launched a $300 million fund to fast-track the development of precision medicine. Machine learning algorithms are incredibly powerful tools that can learn the rules of drug design to improve existing drugs or develop new ones.
Besides increasing the speed and accuracy of treatments, artificial intelligence will play an important role in providing superior patient care. Voice-enabled healthcare assistants like Suki already help lift the burden of medical documentation so that doctors can actually focus on taking care of their patients. Any improvements in misdiagnosis and more efficient drug development will significantly reduce the cost of healthcare.
Check out all areas where AI is challenging doctors on their home turf.
AI and machine learning can be used to automate some of the mundane and repetitive tasks of recruitment and reduce a typically lengthy hiring process.
Startups like Riminder ease the sourcing and screening of job applications with an attempt to surface unusual candidates and prevent recruiting bias.
Textio augments the writing of job posts and direct candidate communications. Unilever even sets up digital interviews to screen entry-level employees using facial recognition and keyword analysis. The latter is a specifically interesting domain with a few companies now entering the playing field, for example Searchie which has a live virtual assistant who interviews candidates and provides a detailed assessment in minutes.
In this increasingly data-driven world the importance and impact of artificial intelligence is only growing and is becoming more crucial in gaining competitive advantage. AI as a service (AIaaS) is helping democratize AI and provides small and large businesses equal access to high quality training data and expertise with less risk and lower cost.
Regardless of the above and all the AI news clogging your inbox, it’s important to understand that while AI might be the future, we should never lose sight of the human touch.
Artificial Narrow Intelligence (ANI)
AI specialized in one field in which it can outperform a human, such as playing a game of chess or Jeopardy or performing surgery.
Artificial General Intelligence (AGI)
A state where machines are equally intelligent to humans in every aspect.
Artificial Super Intelligence (ASI)
A state where AI is vastly superior to the best human brains in every field.
Supervised learning
The machine learning task of training a model on a pre-defined data set consisting of input and desired output values (data labels). This allows an algorithm to provide accurate decisions when new data is provided.
Unsupervised learning
The machine learning task of identifying hidden structures from ‘unlabeled’ data. Examples include clustering (where we try to discover underlying groupings of examples) and anomaly detection (where we try to infer if some examples in a dataset do not conform to some expected pattern).
Reinforcement learning
Learning through trial and error. A computer program interacts with a dynamic environment in which it must perform a certain goal (e.g. driving a vehicle) without a teacher explicitly telling it whether it has come close to its goal.
Neural networks
A type of machine learning model that builds up complex models by connecting simple units (neurons). Each unit is linked to other units in the network and influence the ‘activation’ of these other units.
Training example
A combination of some input data (e.g. the area of a house) and some output data (e.g. the price of a house) that we are trying to predict using a model.
Training set
The set of examples a model is trained on.
Turing test
A test named after computer science pioneer Alan Turing to determine whether a machine is capable of thinking like a human. If an examiner cannot distinguish a response from a machine from that of a human, the test is passed.
Natural Language Processing (NLP)
The ability for a computer or system to understand and process human language as it is spoken. NLP is an area of AI.
Excited to learn more about exponential technology? Interested in becoming a machine learning engineer, data scientist or genomics specialist? Browse our catalogue of curated courses and empower yourself with the necessary tools to advance your career.
The world’s most valuable AI startup with $1.6 billion raised to date specializes in systems that analyze faces and images on a large scale. It is helping China build the world’s biggest facial recognition-enabled surveillance network.
Applying artificial intelligence to develop new medicines for hard to treat diseases. It aims to lower costs, decrease failure rates and increase the speed at which medicines are delivered to patients.
Specialized in developing level 4 autonomous driving technologies to future advanced transportation systems and create a safer and more enjoyable travel experience.
AI-driven music composer that creates & customizes original music for your content.
A fully automated kitchen garden which uses machine learning technology to grow the most nutritious and organic plants without soil.
Backed by Jeff Bezos and Bill Gates, Grail aims to develop a blood test to detect cancer early before symptoms appear.
Two creative experiments from Google: ‘Talk to Books’ is a new way to explore books by starting at the sentence level, rather than the author or topic level. Simply type a question and the tool will return related book sections. ‘Semantris’ is a fun word association game powered by machine learning.
A popular open source machine learning framework build by Google. Playground is TensorFlow’s interactive visualization of neural networks you can play around with to better understand its concepts. Find all of Google’s AI tools here.
A leading open source machine learning and deep learning platform making it easy to build smart applications.
A non-profit AI research company which open sourced tools such as Gym Retro, a platform for reinforcement learning research on games, and Universe, a virtual world for AI to explore and play in. The platform allows for measuring and training an AI’s general intelligence across the world’s supply of games, websites and other applications.
A platform and open source framework launched by the Linux Foundation giving users a visual workflow to design artificial intelligence and machine learning applications, and a marketplace for freely sharing AI solutions and data models.
