Computerized algorithms help us choose which movies to watch, what kind of music to listen to, and what kind of literature to read. Pen, fun, pavilion www. biquge。 info

But what if algorithms themselves transcended the role of human cultural mediums and began to create their own cultures?

In 1950, Alan Turing, a British mathematician and computer scientist, published a paper entitled "Computational Machines and Intelligence" and first proposed a thought experiment called "imitation games". In one room is a human "interrogator", and in another room is a male or female "interlocutor". The goal of the game is for the "interrogator" to determine whether the "interlocutor" hidden in another room is a man or a woman. The parody is done through a series of questions and answers, and the information is sent and received by a third party or by typing. "Winning" the parody game means that the first step in identifying the game has been successful.

Later, Turing modified the parody game by replacing one of the "interlocutors" with a computer, observing whether the computer could successfully complete the conversation, and making it impossible for the "inquirer" to distinguish between a computer and a human "interlocutor". This version of the parody game is called the "Turing Test".

This simple but powerful thought experiment proposed by Turing gives a general AI testing framework capable of studying aspects of the human-machine boundary, and session is just one example.

Writing poetry is compared to artificial intelligence

On May 18, at Dartmouth College, one of the Ivy League universities in the United States, professors from the Department of Computer Science and the Department of Music will discuss different areas of artificial intelligence with students, focusing on the problem of machines creating art. Specifically, in the Turing Test of Artistic Creation, we will observe whether participants can distinguish between sonnets, short stories, and music written by humans and machines. Of course, machine-created art is nowhere near as good as Shakespeare, O'Henry, and silly punk.

Dance music competitions ("Algorhythms") require participants to build an enjoyable collection of dance songs by selecting the most suitable music for use on the dance floor from a preset library. In this case, the computer software selects a random piece of music from a database of dance music as the initial "seed of inspiration" to start the creation. Based on this initial track, the software selects, modifies, and mixes from a library to create a 15-minute dance song. It includes standard annotations for 20 characteristics such as genre, rhythm (BPM), rhythm point, saturation (pitch), and brightness (timbre).

The Sonnet Contest ("PoeTix") and the Short Story Contest ("DigiLit") are even more serious challenges for computers. The contest requires participants to submit a stand-alone software package, based on a specific "seed of inspiration", or enter a common noun phrase (such as "dog" or "cheese grater"), which the software then uses to create the desired literary work. In addition, the software algorithm required to generate an infinite number of different entries from a given prompt.

For testing purposes, we'll first go through the computer's "creations" to exclude those that are clearly machine-based. We mix human and machine creations, and then bring in a panel of "judges" who ask them to tell whether the art is coming from humans or machines. In the dance music competition, we gave it to a group of students to dance to music composed by humans and machines. Works that are statistically indistinguishable from human works will be the ultimate "winners".

This contest is open to anyone. So far, the participants have included academics as well as non-academic practitioners. However, as of now, no company has officially announced its participation. This is a bit of a surprise, given the mushrooming of "machine writing" companies in the literary space, and the increasing use of text-generating software, such as earnings reports and sports event summaries, is becoming more common. Of course, in the world of streaming music playback, many companies are also using artificial intelligence for automated list generation, most notably Pandora.

Judging entries is not straightforward. Even in the original parody game, the gender of the "interlocutor" is revealed over time and revealed through the text. Similarly, in the Turing test, one cannot judge the inhumanity of a computer's utterance from a single interaction implementation, but rather requires a long period of repeated testing.

The background issues of the parody game and the Turing Test are also worth considering. Is the possibility of winning a parody game independent of time, culture, and social class? It is quite possible that in the current Western society, due to the increasingly flexible definition of gender, the original imitation game will be more difficult to win. The same goes for the Turing Test, in the 21st century, we are increasingly communicating with machines. Whether we like it or not, text messaging and instant messaging apps have dramatically changed the way we communicate and what we expect from the communication itself. Abbreviations, spelling mistakes, and omissions of information are almost the norm in people's communication. The same question applies to the field of art.

Where is the line between an artist and a creative aid program? Who is an artist?

Reflections on the art form begs another question: who is the artist? Is the programmer who compiled the sonnet writing software a poet? Is the programmer who compiles the short story writing software a writer? Is the programmer who compiles the music mixing software a DJ?

Where is the line between an artist and a creative aid program? What is the impact of this demarcation on the classification of artistic outputs? The sonnet, a fixed art form, is a high-level algorithm for creative work, albeit performed by humans. Today, when Microsoft's Office software helps you "correct" grammatical errors and "suggests" your choice of words, and you choose to listen to it (whether you're receptive or just lazy), are your creations still your own? Or has it become a "product of human-machine cooperation"?

We look forward to seeing more submissions from "programming artists". Regardless of the performance of these programs, or whether they pass the test or not, their results will continue to push the boundaries of "creation" and "human-machine coevolution".

(by Daniel N. Rockmore and Michael Casey; Professor, Dartmouth College)