Why artificial intelligence is not magic and how to watch it work
For many people, “artificial intelligence” or “machine learning” are terms that sound like magic or modern enchantment, and occasionally a bit threatening. Technologies that take place in the world of computer science, between zeros and ones, and are only understood by the new “wizards”, the Data Scientists. In everyday life, we encounter AI more and more often. Whether at work answering emails, the parallel social worlds when Facebook sprinkles new advertising offers into my feed, or at home on the couch streaming video when Netflix selects just the right movie – artificial intelligence is subtly creeping into our lives. And yet there’s something spooky about the way the AI algorithm seems to know me better and better.
If? Then!
Scoffers like to claim that artificial intelligence is, at the end of the day, just an endlessly nested if/then query that will eventually produce seemingly correct results.
- If a user likes to drink red wine and
- if he likes to travel to France and
- if he has often vacationed in the Bordeaux region and
…
- if he has often vacationed in the Bordeaux region and
- if he likes to travel to France and
- … then he prefers Bordeaux red wine with high probability.
But it is not quite as simple as in this very simple example. Nowadays, artificial intelligence analyzes large amounts of information with the help of artificial neural networks, which ultimately perform mathematical probability calculations and provide weighted result suggestions. The challenge for the human brain is to comprehend these results, which are based on hundreds of parameters. The complexity and speed with which artificial intelligence performs such calculations are not reproducible by humans and are consequently quickly associated with magic.
INTERGATOR SMART Context-Map - A graphical approach
Especially when processing documents and texts with artificial intelligence methods, this comprehension hurdle can be overcome. INTERGATOR SMART SEARCH comes standard with an integrated SMART Context Map that visualizes the relationships between data. Documents with similar content are thematically summarized (clustered) and related to each other. In this way, the user recognizes relevant information around his search terms more quickly and receives an informative advantage over a conventional search. These relationships were calculated by the artificial intelligence behind INTERGATOR SMART SEARCH and are based on innovative methods that understand documents in their entirety and push the presence of contained terms somewhat into the background.
For example, documents may be grouped together that fit together thematically because they have a common umbrella term, but the same vocabulary is not used in either of the two texts. Let’s assume, for example, that you are searching in INTERGATOR SMART SEARCH for “Operating instructions for a drilling machine”. The main features of a drilling machine are “drilling”, “motor”, “drill chuck” and “speed”. In addition to comprehensible search hits such as individual models, drill types or drill catalogs, you will also receive hits for lathes and milling machines within this cluster. Although the latter do not use these terms, they are thematically similar, since a milling machine, for example, also uses a drill-like milling head that is driven by a motor and mills depressions in a surface.
Interaction with the artificial intelligence
Another advantage of the SMART Context Map is the interaction within the graphical interface. The user can pick up the hits within the clusters, using them to refine the original search, give it a new direction, or use it as a starting point for a new search.
With the visualization of the search hits, research is given an explorative component that lifts the magic veil and dispels the idea that AI is a black box whose decisions are not comprehensible. Instead, the SMART context map makes connections clearer and the interaction options allow direct intervention in a previously hidden world.