Celso Bessa (sɛwsʊ bɛ:sa)

Prefiro construir pontes a muros

artificial intelligence

A lot of what is called artificial intelligence is less-stupid-looking automation

por

I was reading an article about the “artificial intelligence” of antagonists in the first The Last of Us, and there is something I believe applies to 90% of what the industry calls artificial intelligence:

“[…] to make the Infected feel grounded, entertaining, and believable.

The best way to achieve these goals is to make our characters not stupid before making them smart. Characters give the illusion of intelligence when they are placed in well-thought-out setups, are responsive to the player, play convincing animations and sounds, and behave in interesting ways.

Botta, Mark. Infected AI in The Last of Us, Game AI Pro 2 – Collected Wisdom of Game AI – edited by Rabin, Steven.

Swap infected and characters for artificial intelligence or the name of an AI product and it applies as well: a lot of what is called artificial intelligence is just automation that acts seemingly less stupid.

Article page with the title "Infected AI in The Last of Us" and below, the text, which highlights the excerpt "The best way to achieve these goals is to make our characters not stupid before making them smart. Characters give the illusion of intelligence when they are placed in well-thought-out setups, are responsive to the player, play convincing animations and sounds, and behave in interesting ways." with the terms infected and characters crossed out and replaced by A.I. (artificial intelligence)
“Artificial Intelligence” is something that is not new in the gaming industry.

It is not an issue for a person or company to spend money on this, even when the benefits are nebulous, uncertain or even non-existent. But when it starts to be uncritically applied in the public civil sphere, including judicial decisions that may be based on fiction, brakes must be put on. Mystification and stupidity, even if well disguised, are acceptable in the private sphere. In public policy and the justice system, it is not.

Coded Bias, an examination of societal issues and artificial intelligence, is available on Netflix

por

Coded Bias (Shalini Kantayya, 2020) is a documentary examining at some of most pressing problems of our age, the nexus of societal issues like racism, genderism, ageism and several biases and machine learning and other forms of artificial intelligence. If you have the opportunity to watch it, please, do it. It is now available on Netflix.

Coded Bias trailer

A few days ago had the opportunity of a live streaming of the movie followed by a conversation Joy Buolamwini (author of Gender Shades, an influential research on racial bias and facial recognition technologies and co-founder of Algorithmic Justice League), Cathy O’neill (author of Weapons of Math Destruction), Ruha Benjamin (author of People’s Science and Race After Technology) and Zeynep Tufekci, and other incredible researchers and activists, in an online event promoted by the African American Policy Forum (AAPF).

Here are some bits of the movie and the conversation that I think are interesting or newcomers to this discussion:

  • The assimetry of the power (“who owns the fucking code”, Buolamwini) in the tech and data police making, gathering and implementation
  • The soundbite “data embbeds the past“
  • How there is no recourse or accountability most of deployed artificial intelligence and biometrics systems and services
  • How powerful Facebook (and other platforms for that matter) are. They could say to regulators or political candidates: “If i churn X voters to you, you don’t regulate me” (A paraphrase of what is actually said by Zeynep Tufekci.
  • Joy asking “who are the gatekeepers of the jobs” of the so-called 4th industrial revolution
  • Joy saying: “people assume if the machine says, it is correct.”
  • Kimberly Crenshaw highlighting how the law protect economic interests and the market by default.
  • Algorithms don’t object or whistleblow when a company,the government or other institution does something wrong [By the way: algorithms, models, automated systems and robots don’t discuss and can’t organize themselves to demand better working conditions and wages]
  • Ruha Benjamin and Joy discussing how there is great power in controlling the labelling and the decisions of using AI and how we need to create/change the language to highlight and reframe societal issues and related use of tech. [Which reminds of how the power of satire and parody for positive society change is understimated by some activists and parts of the academia]
  • Crenshaw drawing the attention to the importance of fighting for tech literacy and real accountability

As mentioned, If you have the opportunity to watch it, please, do it. It is now available on Netflix. And if you are interested in setting up a watch party and discuss the movie, let me know.

a website showing up a digital white face with some geometric shapes over it and the text about doing a watch party
Watch Party pages at the Coded Bias movie website

We are so much more than what they are forcing us to accept.

por

We think intelligence is about playing Go. But practically no one plays Go. Or chess. Real intelligence is being able to walk through an incredibly crowded street on a busy evening, nimbly, when you don’t even think about it, while at the same time recalling memories and replaying things in your brain. What I’m saying is that human beings have been reduced to a very simplified version of themselves, which they’ve accepted, in order to fit into this machine model, both of society and the internet. But we are extraordinary and we can do extraordinary things. We are so much more than what they are forcing us to accept.

Adam Curtis interview to Tom Hodgkinson at Idler

Reading: AI ethics groups are repeating one of society’s classic mistakes

por

Too many councils and advisory boards still consist mostly of people based in Europe or the United States […] This lack of regional diversity reflects the current concentration of AI research [artificial intelligence]: 86% of papers published at AI conferences in 2018 were attributed to authors in East Asia, North America, or Europe. And fewer than 10% of references listed in AI papers published in these regions are to papers from another region. Patents are also highly concentrated: 51% of AI patents published in 2018 were attributed to North America.

Source: https://www.technologyreview.com/2020/09/14/1008323/ai-ethics-representation-artificial-intelligence-opinion/?itm_source=parsely-api