Artificial Intelligence: “Soft vs Hard”

Jillian Holbrook
HackGuild
Published in
3 min readJul 21, 2020

Artificial intelligence (AI). When first mentioned, AI may bring to mind visuals of supercomputers and robots taking over the world — like, for example, Avengers: Age of Ultron. However, there are different qualifications that can be made about the extent of the autonomy of AI. Learning what types of AI are out there can help us learn whether we need to fear a technological apocalypse with robot overlords.

What is Artificial Intelligence?

To understand the different types of AI, it’s important to know that artificial intelligence is the ability for machines or computer systems to complete a task typically requiring human intelligence through the use of advanced algorithms.

These algorithms allow technology to aid in translation for foreign language, as with Google Translate, recognize speech for purposes like checking the weather or turning on the lights, or even to make decisions!

“Soft” Artificial Intelligence

Artificial intelligence classified as “soft” or “weak” is response-based AI. In other words, the technology is not actively thinking for itself.

Common “soft” artificial intelligence systems may be as nearby as your pocket! Personal assistants like Apple’s Siri and Alexa from Amazon are excellent examples of “soft” AI.

For instance, imagine getting ready for the day and wondering what attire is appropriate. If you were to own some sort of device with a technological assistant, you could simply ask, “What’s the weather like today?”

The personal assistant technology is not actively thinking about what the weather outside is like. Instead, it is using keywords and phrases through speech recognition algorithms to provide a response. The words “weather” and “today” can be used to prompt the technology to use location data and information from meteorology services to tell you whether or not you need an umbrella, a jacket, snow boots, or some strong SPF sunscreen.

Therefore, we can think of “soft” or “weak” AI as an input-output system. Information goes in, the input data is processed, the technology uses algorithms to complete a desired task or function, and finally, an action or answer is then spit back out. Yet the technology is not using the information to learn and get smarter.

“Hard” Artificial Intelligence

“Soft” artificial intelligence doesn’t seem so scary. It obeys our commands with no mind of its own. However, the other classification for AI, “hard” artificial intelligence is a slightly different story.

The most prominent difference between “soft” and “hard” AI is that “hard” AI does not just take in information — it actively works to comprehend the information and carry out tasks with its own volition. Where “soft” AI is predictable, “hard” AI is more like the human brain itself. A technological entity with the ability to think and process for itself is more representative of human intelligence and action than algorithms that prompt information and task spitballing.

Some of the best examples of “hard” AI are actually robots that have been taught to play poker or video games. By examining the variables of a game mathematically, they can develop their own strategies. This association-style processing allows the technology to learn instead of stagnating in the ask and answer style of “soft” AI.

Looking Forward

In the end, how we look at artificial intelligence is all about perspective and classification. While we don’t need to constantly be fearing superintelligence evolution from our technology, it is important to understand the impacts that AI may have in the future. Automation and technology do have the potential of simplifying our lives, but evaluating ethics in technological fields, including AI, is important as we continue to develop smarter devices.

Article By: Jillian Holbrook

--

--