Generative AI: Don’t call it ‘artist,’ say scholars in Science magazine

It’s no secret that people’s assumptions about artificial intelligence are generally ridiculous. They routinely allege a presence that is really just a machine, much as former Google engineer Blake Lemoine claimed last year a chatbot, LaMDA, was “sentimental”.

Even journalists tasked with conveying an understanding of technology often pursue propaganda rather than knowledge.

Now, in an effort to counter some bad perceptions and focus attention on important AI questions, a group of scholars warned Thursday that humans should rid themselves of the misconception that an AI program is the work of an “artist.” Let’s constitute.

Too: What is AI? Everything To Know About Artificial Intelligence

“The term ‘artificial intelligence’ can misleadingly imply that these systems display human-like intent, agency, or even self-awareness,” wrote Ziv Epstein, a post-doctoral researcher at the Massachusetts Institute of Technology’s Media Lab, and several colleagues. Are.” In an article in the prestigious journal, Collaborating Institutes Science,

The article, “The Art and Science of Generative AI,” was published online Thursday for the June 16 print edition of Science.

The authors write, “Natural language-based interfaces are now accompanied by generative AI models, including chat interfaces that use the pronoun ‘I’, which can give users a human-like interaction and a sense of agency.”

The authors caution that attributing agency to an AI program may lead people to adhere to a number of false assumptions, such as failing to attribute the human creator of the work.

The authors write, “These assumptions may reduce credit to the creators whose labor reduces the system’s output and deflect responsibility from developers and decision-makers.”

The article addresses several viewpoints about AI that have become prevalent in recent years.

Too: Best AI Chatbots: ChatGPT and Other Notable Alternatives

In recent years there have been some infamous art-world demonstrations of “AI-generated art” that, in fact, obscure the origins of a work’s production.

One of the brightest PR moments of the AI ​​\u200b\u200bera was the sale by Christie’s auction house in October 2018A painting output by an algorithm named “Edmond D Bellamy” for $432,000. The painting was introduced by auctioneers and curators as “created by artificial intelligence”.

At the same time, however, a thread of research has spanned many years in which computer scientists and artists have tried to fashion artworks with computers that can produce quasi-random effects. That thread of inquiry seeks to discover the dividing line between human control and computer spontaneous production that escapes human control.

Such programs are often grouped as “generative art systems with intentionally unpredictable outputs”, or sometimes referred to as “decentralized autonomous artists” or “autonomous art systems”, Epstein told ZDNET. Mentioned in an email.

Too: These are my 5 favorite AI tools for work

Epstein said, examples of productive programs include boto, electric sheep, Abraham, meet ganimalAnd artificial fm,

Some programs use so-called “generative AI”, referring to such generative adversarial networks, or “GANs”, including Boto and Ganimal. But not all programs use that approach.

Epstein told ZDNET, “As opposed to direct input-output use of generators, these often involve systems that respond to input from a crowd, or some other data stream, and thus create outputs that are intentionally unpredictable.” Are.”

The generative movement has explored the line between human intent and spontaneous machine output.

In 2006, long before the current wave of AI, computer science professor Simon Colton K. Queen Mary College, London, offered what he called “The Painting Fool”, a program that would produce images through a mix of rules given by a human operator, but also other parameters the human did not explicitly manipulate. Detecting a person’s “emotion” through emotion-recognition software, for example, would use the interpretation of that emotion as an input for a work of painterly art.

Too: Best AI art generator: DAL-E2 and other fun options to try

in The Guardian newspaper in 2017Colton told reporter Gemma Cappala-Ramasamy that he aimed to “explore what it means for software to be creative”, with the aim of “one day being taken seriously in software as a creative artist in its own right”.

In recent examples, artists have drawn attention to things they found in the generative AI’s output, which they noted were surprising, or that something was going on with the program beyond their expectation, Like last year’s recurring figure of Loab.

In Science, Epstein and team refer to a very long article, Posted on arXiv pre-print serverIn which he elaborates on how it is important not to carelessly treat a machine as autonomous, no matter how impressive the consequences.

Epstein and team emphasize the option of discussing generative AI “as a tool to support human creators, rather than as an agent capable of harboring its own intention or authorship. In this view, autonomous There is little room for machines to be ‘performers’ or ‘creators’ on their own.”

Too: How to use the DALL-E 2 to turn your ideas into AI-generated art

The authors base their point on a concept called “meaningful human control” or MHC, which, they write, is
“Basically adapted from the autonomous weapons literature refers to the control and responsibility of a human operator for a computational system.”

“MHC is achieved,” they explain, “if human creators can creatively express themselves through the generative system, producing an outcome that aligns with their intentions and their personal, expressive signature.” carries.”

Future work is needed, Epstein and his colleagues conclude, to find generative systems that let human users perform “fine-grained manipulation on output” while using AI to do the work.