Article Author:
Canadian Press
THE CONVERSATION
This article was originally published in The Conversation, a source of independent, non-profit news, analysis and commentary from academic experts. rice field. Disclosure information is available on the original site.
-
Author: Jordan Richard Schoenherr, Assistant Professor of Psychology, Concordia University
After Blake Lemoine, active activity on social media Occurred Google developers took a vacation claiming that the chatbot LaMDA became sensuous, that is, gained the ability to experience emotions. Lemoine, in support of his claim, posted an excerpt from his interaction with LaMDA. LaMDA answered the question, "I want to be aware of my existence and want to know more about the world. Sometimes I feel happy and sad." He also states that he has "the same desires and needs as people."
It may seem like a trivial exchange, and while it seems more realistic than the initial attempts, it is hardly worth claiming sensibility. Even evidence of the Lemoine exchange was edited from several chat sessions. Nevertheless, the dynamic and fluid nature of the conversation is impressive.
Before creating the Artificial Intelligence Rights Code, we need to consider how human experience and prejudice affect our trust in artificial intelligence (AI).
Creating Artificial Intelligence
In popular science, AI has become a comprehensive term and is often used less often. Artificiality emphasizes the non-biological nature of these systems and the abstract nature of the code, as well as the inhuman pathways of learning, decision-making, and behavior.
By focusing on man-made, we can overlook the obvious fact that AI is created by humans to make and support human decisions. The consequences of these decisions can have a consequential impact on humans, including credit judgment, spouse discovery and selection, and even potential criminal decisions.
Chatbots (great ones) are designed to simulate human social interactions. Chatbots have become a very familiar feature of online customer service. If your customers only need predictable responses, you may not know they are interacting with AI.
Function of complexity
The difference between a simple customer service chatbot and a more sophisticated type like LaMDA is the data used for AI training. Rules to manage sets and exchanges.
Intelligence reflects several features — there is domain-specific and domain-general form of intelligence. Domain-specific intelligence includes tasks such as biking, performing surgery, naming birds, and playing chess. Domain general intelligence includes common skills such as creativity, reasoning, and problem solving.
Programmers have come a long way in designing AI that can demonstrate domain-specific intelligence, from performing online searches and playing chess to recognizing objects and diagnosing medical conditions. If you can determine the rules that govern humans, you can think and teach AI those rules.
General intelligence (what many people typically consider to be human) is a much more complex ability. For humans, we may rely on the confluence of different types of knowledge and skills. Language-like features provide a particularly useful tool that allows humans to remember and combine information across domains.
Therefore, developers have often expected the potential of human-like artificial intelligence, but these hopes have not yet been fulfilled.
Beware of AI
AI claims that it may be a sensory current task that goes beyond the general intelligence task. increase. Philosophers have long pointed out that it is difficult to understand the mental state of others, not to mention what constitutes the consciousness of animals other than humans.
To understand the sensibility claim, we need to look at how humans judge others. We often misattribut our actions to others and assume they share our values and preferences. Psychologists say that children need to learn about the mental state of others and can improve their ability to understand others if they have more models or are incorporated into a more groupist culture. I observed.
When judging the intelligence of an AI, it is more likely that humans are anthropomorphic than the AI actually has perceptual power. Much of this has to do with familiarity — you can increase your preference for objects and people by increasing their exposure to them.
Sensitivity claims by people like Lemoine should be interpreted from this perspective.
Can you trust the AI?
The Turing test can be used to determine if a machine can think in a way that is indistinguishable from humans. LaMDA's response is certainly human-like, which means it's good at learning patterns. No sensitivity is required.
Just because someone trusts a chatbot does not mean that trust is guaranteed. Instead of focusing on the highly speculative nature of AI's sensibilities, we need to focus on efforts to address social and ethical issues that affect humans.
We are faced with the digital divide and the power and distribution imbalance between what we have and what we do not have in the creation of these systems.
The system must be transparent and accountable so that the user can decide. Explainability requires the cooperation of individuals, governments, and the private sector to understand and regulate artificial intelligence and its applications.
It should also be noted that the tendency of humans to anthropomorphize can be easily abused by designers. Or they may reject useful AI products that are not human. In an era of entanglement, we must be critical of who we trust and what we trust.
-
Jordan Richard Schoenhar works, consults, owns shares, and receives funds for companies and organizations that benefit from this article. It is not. Academic appointment.
-
This article has been republished by The Conversation under a Creative Commons license. Disclosure information is available on the original site. Read the original article:
https://theconversation.com/rather-than-focus-on-the-speculative-rig https://theconversation.com/rather-than-focus-on-the-spec