Did you realize something? When analysts and media write about artificial intelligence (AI), most of them unfortunately only talk about machine learning. In doing so, they mention AI and machine learning in the same breath and thus equal AI with one single technology. This is wrong and a concerning progress. In particular, it is confusing the market during a time when 58 percent of organizations worldwide (according to Forrester) are still researching AI. However, AI is more than just machine learning and consists of several different components that provide intelligent solutions.
Machine learning is not equal to AI
First of all, machines do not understand. This is by far the biggest misconception while discussing AI, in particular in the context of virtual private assistants like Amazon Alexa or Apple Siri. Machines match data to predefined data patterns of understanding. Thus, understanding is a question of the size of a data pool, because the more data is matched to something we can understand the more “understanding” a machine seems to have.
The biggest issue is that AI research has been an oscillating system between several techniques. Whenever one does not do “the job completely,” people get frustrated and turn to another one. And right now, the market is of the opinion that organizations do have tons of data that can be utilized together with machine learning algorithms. And since some machine learning use cases succeeded, machine learning is hyped by the media.
However, machine learning just helps to identify patterns within data sets and thus tries to make predictions based on existing data. It is most important to check the plausibility and correctness of the results since you can always find something in endless sets of data. And that’s also one of the drawbacks if you consider machine learning as a single concept. Machine learning needs lots of sample data or data in general to learn and be able to find valuable information respectively results in patterns.
Facebook’s News Feed is a good example for machine learning to personalize each member’s feed. Meaning, a member who frequently stops scrolling to read or like a certain post of a friend will see more of that friend’s activity.
However, our thinking patterns are highly complex and composed of many techniques while we are making a decision. Machine learning is one component of an AI, not the AI.
Artificial intelligence needs perception and problem solving
Indeed, machine learning is an essential part of AI. However, it is just one single part of the entire ecosystem that is needed to build and run a sophisticated AI inside an organization. With that being said, an AI consists of two major parts: “perception” and “problem solving.”
- The perception part (aka data view) represents all the necessary input streams like sensors (equivalent to using eyes to see, ears to hear or hands to feel) but also the needed skills (algorithms) to pattern match as well as to utilize the learned patterns that can be used to execute a task like “find a dog with long curly fur inside a pool of pictures of cats, dogs and dust mops.” This is typically an area where machine learning plays the major role right now.
- The problem solving part (also known as decision view) is responsible for the actual human-like intelligence inside an AI. It represents the human-related experience like e.g. logical arguments and evaluation procedures that combine rational thoughts as well as emotional decisions (gut feeling), something a machine is not able to develop on its own. That part of the AI provides the contextual based decision-making process. A characteristic, machine learning algorithms again are not capable to deliver.
Bottom line, an AI needs the ability to collect, conjunct and analyze data inside and across large pools of data sets. That is the ability machine learning algorithms provide us with i.e. pattern matching. But what’s next? How is a system going to decide after a successful match? And how does a machine decide, in particular, when it faces a dynamic environment? That is the specific area where a machine is taught human-like behavior and is empowered with the capability of learning on its own and find solutions without human interaction. Here an AI needs the ability to utilize existing knowledge, context and experience to run a decision-making process.