Artificial Intelligence – Human-like or Not?

Artificial intelligence has captured the world, especially the tech community, with great excitement and fear. Despite its fame, there lacks a universal agreement on what exactly the phrase means.

In class, Head of Science and Engineering at LinkedIn Craig Martell distilled what artificial intelligence is. He emphasized that the buzz phrase is a misnomer given that there is no magic that will allow machines to gain consciousness nor anything that resembles human intelligence. Rather, artificial intelligence and machine learning ultimately comes down to probability calculations in higher dimensions that enable the machines to output the most likely outcomes. In other words, AI, at its core, involves complex statistical models.

SAS describes artificial intelligence in a similar tone, emphasizing that machines take in data to make adjustments to its algorithm in order to perform tasks. However, it shares a definition that incorporates more analogies to the human mind than Martell. For example, SAS mentions that machines “learn from experience…[to] perform human-like tasks” [1]. This diverges from Martell’s emphasis that machines can now perform tasks that even humans could never do given that they are able to make calculations in much larger dimensions than the human mind was ever capable of grasping.

On a further extension, BBC provides an explanation of AI that is even less aligned to Martell’s definition but similar to what people may expect in science fiction. BBC says that AI involves “creating a computer mind that can think like a human” and that by 2050, computers will essentially be able to act like humans, if the technology does advance to that stage [2]. Investopedia also adds a more human-like quality to its definition that “machines are programmed to ‘think’ like a human and mimic the way a person acts” [3].

Overall, all three definitions align in that there is some computer that takes in data as input and returns an output that completes some task. However, there is a clear range in the representation of how human-like the machines can get in terms of intelligence. With the perspective of a computer scientist, Martell’s definition is most accurate given that AI truly does come down to lines of code that reads in data, updates some statistical model, and returns a calculated value that represents the most likely outcome. Amidst all this coding, there is no piece that enables the machine to have any other form of human characteristic other than doing math far superior to the average human brain. Perhaps the connection of AI mimicking human behavior has any combination of the following reasons:

  1. To help nontechnical people understand AI as some black box that makes decisions by learning from data,
  2. To create a more exciting storyline to increase the hype,
  3. To alleviate the actual difficulty and mundaneness of the work behind AI and create a movement that can sweep up the excitement and attention of all people.

Whatever the reason may be, it’s good to keep the true core idea of what artificial intelligence truly is as Martell revealed. Simultaneously, it’s also quite interesting and possibly exciting to think about how certain parts of science fiction might become true in ways that will positively impact the world.

Sources:

[1] – https://www.sas.com/en_us/insights/analytics/what-is-artificial-intelligence.html

[2] – https://www.bbc.com/news/av/technology-34224406/what-is-artificial-intelligence

[3] – https://www.investopedia.com/terms/a/artificial-intelligence-ai.asp

0

2 comments on “Artificial Intelligence – Human-like or Not?”

  1. Hi spak7! Great discussion on the definition of AI.

    I definitely agree that Mr. Martell’s definition of AI is the pure definition – making predictions based on statistical models. However, I think we should address humans influence on AI. The history of AI, starting roughly in the early 1950s, started off with a human comparison – the Turing Test. If a computer is indistinguishable from a human, then the computer is considered intelligent or as having the capability to “think.”[1] In the 1960s, computers were programmed to communicate in natural languages, such as English. [1] The Robotics boom, starting in the 1980s, talked about robots needing to have bodies to perceive the world. [1]

    The language we use to talk about machines is so heavily tied to how we talk about ourselves as humans. I think it’s important to be aware of the interconnectedness of the AI and humans, and not shy away from those implications. Yes, at the moment AI is statistical models, but also it could be so much more.

    [1] https://en.wikipedia.org/wiki/History_of_artificial_intelligence#Turing's_test

    0
  2. Craig Martell’s definition of AI resonated with me. I think there is much more to intelligence than just processing speed. Computers need to be “self-aware” to come close to human intelligence. Humans have the gift of being able to make conclusions based on observations. While computers are making progress in the area of pattern recognition, they still need to be fed what patterns to “look for”.

    0

Comments are closed.