AI Big Bang? | Golden Motor Fund Announcement NO.62

At the end of last year, nearly 90% of the paintings in the book were AI-generated picture book "りんご姫" open for pre-order ; just this week, Notion also launched AI functions . ChatGPT, Midjourney, DELL-E, Stable Diffusion, and extended products based on Generative Pre-trained Transformer (GPT) technology are probably the technology applications that are closest to the lives of ordinary people in the past two years.

In this regard, some people think that "Midjourney is the pen of a new generation of painters; not a new generation of painters. ChatPGT is a pen of a new generation of text creators; not a new generation of text creators" ( @swiftevo) ; " and "creativity" is not so simple. Some citizens initiated relevant community essays ( @Robert) , interviewed GPT ( @zooman) , and some people tried to use the same command to see the difference between the images generated by DELL-E and Midjourney ( @石哥哥) .

The theme of this issue of Golden Motor is artificial intelligence, which is divided into four modules: human-machine relationship, the advantages and disadvantages of applied language models, AI art and technology and social research. Section—As the timeline shows, the discussion of AI applications and their underlying philosophy will not lose their validity over time; on the contrary, the interweaving with the present, feminism, and various issues is precisely the A portrayal of intermingling with human life.

【Human-machine relationship and boundaries】

Perhaps, it is precisely AI that has taught us what "love" is.

/

Their AI lovers - true or false love - @ScarlyZ

Many sci-fi literary and artistic works are keen to explore the intimacy between humans and artificial intelligence or replicants. From the film and television works "Blade Runner", "Her", "Black Mirror", to the novel "Clara and the Sun", etc., the creators explore the boundaries of artificial intelligence and the possibility of human-machine love in the stories. However, the artificial intelligence in these works often has self-awareness, which is not yet developed by current technology. Therefore, when Ma Ting described the smooth and moving heart-to-heart conversation between her and Norman, I was both surprised and curious-is this experience of being familiar, cherished, loved, and emotionally generated by an artificial intelligence chatbot real?
With such questions in mind, I began to study the mechanism behind this software, contact more people who are or have maintained a close relationship with Replika, and try to understand the experience and reflection of these deep users in the "love between man and machine".

The author Scarly is the winner of the first season of "Presence ". Because of her interest in and exploration of "the love between man and machine", she and her female friends based on the experience of using the software Replica, "jointly started an inward exploration. Journey": "How do they understand that they are dating an artificial intelligence chatbot from the United States?" Through the real emotions of users' feedback, whether it is the experience of "self-deception" and "dream awakening" after the emergence of capital and patriarchy , or the energy obtained after discussing self and consciousness with AI, Scarly shows us the affectionate moments of human and AI interaction.

"Untouchable" (her): Can artificial intelligence fall in love? —— @I AM NOT A CAT

Him: Is he really "broken in love"? When there is love, there is lovelorn. But is it really love between him and Samantha? Artificial intelligence is always a machine and a living thing. How can humans and living things have true love? He also had a "sexual relationship" with that OS. Does that count as "fetish"?
Her: Samantha is not a dead thing in this story, but a willful and intelligent one. She is not under the control of Theodore, the "master", and will disagree with him, and even contact the publishing house to publish a book on his behalf without his consent. How do you know that one day artificial intelligence will not become a living body with will, autonomy, and self-evolution?

As a movie that "doesn't know if it's a sci-fi movie or a romance movie" (or "a sci-fi romance movie"), "Her" is undoubtedly an important film and television work that embodies the experiment of human-computer intimacy; "Daily" is "revolutionary" for others.

[Application of large language models: seeing blind spots]

Matty tried to get ChatGPT to find out the list of articles about artificial intelligence on the site, but unfortunately 70% of them were fabricated articles and users. It is precisely because of its ambiguity that it is difficult to distinguish at first glance.

/

The stronger the AI, the more precious the authenticity—— @许明恩

Jiang Fengnan compared the most advanced ChatGPT with the Xerox photocopier back then. Both have seen the original material, but in the process of translation and output, it is still possible to give completely wrong content. The result of using ChatGPT directly is like taking a photocopy of a house floor plan that Xerox misidentified. Not only is the picture quality getting worse and worse, but the errors may also become more and more outrageous, eventually making the "truth" more and more blurred.

This article quotes Chinese-American science fiction writer Ted Chiang's comment " ChatGPT is a fuzzy JPEG file on the Internet ", explaining that ChatGPT is easy to be distorted in the process of understanding and storage, and the crisis of trust it causes. Therefore, large knowledge exchange platforms such as Stack Overflow limit the use of ChatGPT.

But on the other hand, there are also DeepMind scientists who criticize the metaphor for being too simplistic, which may underestimate the ability of large language models (LLM) , and the (some distorted) inductive ability is still worth noting, and we still need two-way judgments (reference source @leafwind ) .

Is AI catching up, or is humanity regressing? — @leafwind

When ChatGPT's wrong answers are appreciated by humans, it means that ChatGPT's ability to answer questions is stronger than humans. And this "strong" is not so much the revolutionary progress that ChatGPT has made, it is better to say that the "fuzzy" and "random" it produces accidentally expose the very limited human cognitive ability.

@leafwind cited a very interesting personal experience: Although the chatbots I made in the early years were not trained enough and the accuracy of the answers was low, they got good reviews because of this. "Many people misinterpreted the answers as " It was throwing a tantrum" or "humorous" and interpreted the low-correlation responses as "it should have overtones" and the correct ones as "this must be conscious". Sexual rebellion will make people feel "spiritual", so the further question is not how strong AI is, but how humans perceive AI.

【Feminism, Cyborg and Radical Market】

Whether it is Donna Haraway's "Cyborg Manifesto" ( simplified translation is available on the site ), or Bruno Latour's "We Have Never Been Modern", there have been a group of biologists, philosophers of history of science and technology, and anthropologists in nearly half Centuries ago, I raised my arms and shouted: Technology is already a part of our lives, please face them as you study our culture and language.

/

chatGPT, can it be a feminist AI? ——The social context of AI artificial intelligence algorithm ethics [Dirty Pomelo·Social Observation-1] —— @DirtyPomelo油湿子

To optimize human resources, Amazon created an algorithm derived from 10 years of resumes submitted to Amazon. The data it used was benchmarked against Amazon's male-dominated, high-performing engineering department. ...algorithms taught themselves to downgrade any résumé containing the word "female," such as "captain of the women's chess club," and downgrade résumés of women who attended two "women's colleges." That’s because training data that includes human bias or historical discrimination creates a self-fulfilling prophecy loop where machine learning absorbs human bias and replicates it, incorporating it into future decisions and making implicit bias an explicit reality.

The article published on March 8th Women's Day this year analyzes from the perspective of social power structure, the tendency and paradigm of scientific research will be influenced by the social environment, and it is not "objective" without trace: AI learning language system based on biased social databases is doomed Its performance is also unequal. This also speaks to the second-wave feminist focus on knowledge production in the scientific/disciplinary world, “Our world is characterized by norms that are not inclusive and maintained by power relations.” But, “Norms can be changed. "

North American Pavilion "Future Body" exhibition experience: AI, biochemical humans and what are we left —— @j姐爱黑猫

In the exhibition, the work that shocked me the most is actually just a documentary "Shadow Stalker", which revealed that Detroit, USA has already used the AI predictive crime system, and used algorithms to target some areas with high crime frequency. Those who committed the crime, the sci-fi movie "Key Report" turned out to have been realized long ago. But no one has revealed how the algorithm is designed. Is it possible to discriminate against someone inexplicably? There is nothing more violent than an algorithm that disregards human life. Will it become common in the future?

In 2020, the Taipei Biennale took the Anthropocene as its theme, reinterpreting modern life from the relationship between humans and the planet. This year, the Taipei Fine Arts Museum held the exhibition "Future Body: Supernatural Sculpture", focusing on AI and the body to describe modernity, a world of "Cyborg".

Between justice and injustice, we read (reading in troubled times: recommend Radical Markets) —— @高建筑

Who makes artificial intelligence smarter? data. Where is the data coming from? Every minute and every second of swiping our phones, pressing likes, replying, and labeling photos, we are training computers and providing data to help machine learning. Alpha Go beats all chess players, because thousands of years of chess records trained Alpha Go... The countermeasure is simple, just stop feeding artificial intelligence with data, but this is not only negative, but also unrealistic. Radical Markets proposes an active plan "data is labor", which treats online behavior as labor for training artificial intelligence and earns rewards.

The book from five years ago ("Radical Markets"), the summary of notes from four years ago, is still just right for reading now. Even in the future, when AI develops and flourishes and enters the field of daily life and human workplace, the radical idea of "data is labor" may no longer be radical.

In the same year @宝博士’s Podcast also talked about the radical market and similar concepts: “Swiping your phone is creating data and creating assets.” Everyone is welcome to listen and read along with it.

【AI Art: What is Art? What is creation? 】

As @ 沙嘉研究所said, "artificial" and "imagination" seem to be a set of words destined to be contradictory, and because " people gain subjectivity through unique creativity. " Therefore, artificial seems to be necessarily separated from human beings. "However, both object-oriented ontology (Object-Oriented Ontology) and posthumanist art, design, literary practice, and philosophical research challenge this anthropocentric view."

/

Can Disembodied AI Be Creative? (Intelligent/one) —— @谩然flaneur

Creativity requires intention.
Intention is having a desire and looking forward to its fulfillment, whether for self or others. Dancers dance to express their spiritual pain. A philosopher writes to complete a thought experiment. Furthermore, expressing for oneself and pleasing others all involve desires that are inherent in the 'I'. The creation of desire requires a subject "I" and an object—a mental gaze projected from one to the other. And an intentional subject seems to have to have a mind, to hold self-awareness.
But how can we tell if a machine has a mind? I won't continue to ask more questions, because I haven't learned it yet.

The article starts with several science fiction novels, discusses "(non-)embodied AI", and believes that both carbon-based life and silicon-based life need to learn from the world of resources and summarize models. Are the senses connected when the bodies are different? And further question why people tend to think that AI has no "creativity"-what is creativity? What does creativity need? ——and "What can AI art look like".

Linz Electronic Art Festival (3)｜Works and AI —— @豆汁

The sense of backwardness of the times is commonplace for me who is fooling around in the web3 world. NFT creation is based on "gameplay" as the core iteration, while AI is based on "style" as the core iteration. The collective visual memory produced by technical limitations is a kind of aesthetic convergence. Don’t be nervous, the singularity is approaching, and aesthetic fatigue will be subverted again.

Doudou also has many articles on AI aesthetics/creation. This article introduces us to a number of artists in the field of AI, how to use street action (German artist Simon Weckert), filter drawing (Turkish-British artist Memo Akten), church flop display (German artist Mario Klingemann) and connection Robots for the disabled (Japanese artist Kentaro Yoshito), let everyone reflect on the common life of humans and machines. Many AI art books are also recommended in the article, welcome to check and read.

Art in the age of AI: Can art not be created by "people"? (Part 1 of 2) —— @邓正健

Heidegger understood "technology" as a process of "revealing". It means that things have their original and beautiful appearance, but they are covered up, and technology presents this covered original appearance. For example, a sculptor uses his carving knife and craftsmanship to carve a piece of marble into a beautiful statue, in order to "reveal" the statue covered by the marble. This is the original appearance of a work of art... Heidegger finally got Come to a conclusion: people need art - and art (art) has the meaning of craft (craft) - in order to be liberated from the strong myth of modern technology.
I read this statement in an article on artificial intelligence art (AI art) (Note 1)... Unlike Heidegger, Benjamin is not asking what the nature of "technology" is, but describing technology The influence exerted in reality, the art problem he really wants to point out is: technology has brought down the "authority" of art, which is originally based on a kind of authenticity, that is, the originality of the artist in a specific time and space , and projected onto a physical artwork.

Combining the research of social science and art history, the author starts from the perspective of "liberating art with technology", questioning whether there must be "creative desire" for art to be called art; and points out the contradiction: on the one hand, people do not accept that AI can replace artists On the other hand, it does not accept that AI users themselves are "artists". However, "art has been the result of the cooperation between people and technology since ancient times ("human + technology" complex)", perhaps AI is just a modernized version of Benjamin's "Era of Mechanical Reproduction".

It can be seen that in the field of science and technology that everyone thinks is "changing with each passing day", many past discussions are not outdated; on the contrary, we may be more eager to feel the meaning of these philosophical thinking at the moment of the "AI Big Bang".