“x uses Artificial Intelligence to…” — it probably does not

“Facebook uses artificial intelligence (AI) to customize your feed.” “Microscope uses artificial intelligence to find cancer cells.” “Google Calendar uses artificial intelligence to help people schedule in time.” Or does it?

The term is vague. On Wikipedia, we will find broad definition that let us label many algorithm types as AI:

Artificial intelligence (AI) is the intelligence exhibited by machines or software. It is also the name of the academic field of study which studies how to create computers and computer software that are capable of intelligent behavior. Major AI researchers and textbooks define this field as “the study and design of intelligent agents”, in which an intelligent agent is a system that perceives its environment and takes actions that maximize its chances of success.

So yes, you may be technically correct using this term in a context of bots, speech-to-text, image feature extraction, song recognition, data mining or decision support.

There are better and more meaningful words you can use, however. If you’re a tech journalist, do your homework and instead of giving terribly generic “artificial intelligence” phrase in your publication title, try something more precise. For starters, try replacing “artificial intelligence” with “magic”. If it makes as much sense and carries as much informational value as before the replacement — you failed the test.

There are many terms that are much more meaningful, but it is still important to explain them properly. Below — my short unscientific “definitions” of some important topics connected with AI.

• Artificial neural networks — ANN are computational models inspired by biological neural networks, where you have multiple (often unknown) input signals and the job is to process it to obtain useful information. ANN can be both digital and analog (electric circuit). Main elements of neural nets are neurons (nodes) and connections between them. Input signals propagate trough layer(s) of neurons being subjected to mathematical or signal operations like multiplying, aggregating, adding, subtracting, etc… The basic premise is to use abstract and generic structures — universal ANN layouts for a lot of different tasks like image recognition, speech-to-text or clustering. Basically everywhere where computational problems are hard to solve using ordinary rule-based programming. So how you choose the neural nets parameters and connection weights? That’s where “magic” happens. You teach ANN using curated sample data. That’s how you set all the parameters to fixed values. ANN with fixed params/weights can be used for actual tasks, where the output signal from ANN is the “result” of desired computation.
• Machine learning — ML is a field of science devoted to creating algorithms that can “learn” and adapt based on the processed data set (they improves with time). Such algorithms operate on the basis of a dynamic model formation, based on input signals (data), and use it to create conclusions, decisions and knowledge. Machine learning promises construction of algorithms that can learn from and make predictions on data. Most common applications of ML are spam filters, recognition and image processing, weather forecasts or search engines.
• Natural language processing — NLP is concerned with the interactions between computers and human natural languages in contrast to formal languages communication (such as programming languages) or predefined commands. NLP may include some pre-processing like OCR or speech-to-text to prepare proper text input for further transformations.

This short list is neither comprehensive nor sufficient. Above terms are more specific than “AI”, but still represent a huge and broad scientific research areas. “Machine learning” is another buzz phrase these days, actually. And the sad thing is, tech companies usually include quite a lot valuable technical information in their press materials or on their websites. Virtually no one uses it in articles, however.

Next time when you will be tempted to use “artificial intelligence” in your title, think about “magic” replacement. It can do miracles!