Tech firms already hail AI language models that can interpret or generate text as a breakthrough, but research shows that these models exhibit a troubling tendency toward stereotyping, toxicity, and racism. (From 2021).
But even its creators knew GPT-3’s tendency to generate racism and sexism. Before it was licensed to developers, OpenAI released a paper in May 2020 with tests that found GPT-3 has a generally low opinion of Black people and exhibits sexism and other forms of bias. Despite those findings, OpenAI announced plans toa month later. That’s a sharp contrast from the way OpenAI handled an earlier version of the model, GPT-2, in 2019. Then, it initially released only small versions of the model.
Text generated by large language models is coming ever closer to language that looks or sounds like it came from a human, yet it still fails to understand things requiring reasoning that almost all people understand. In other words, as some researchers put it, this AI is a fantastic bullshitter, capable of convincing both AI researchers and other people that the machine understands the words it generates.
“The definition of bullshitting is you talk a lot and it kind of sounds plausible, but there's no common sense behind it,” Gopnik says. Choi is also working on ways to reduce the toxicity of language models. Earlier this month, she and colleagues introducedthat learns from offensive text, similar to the approach taken by Facebook AI Research; they say it reduces toxicity better than several existing techniques. Large language models can be toxic because of humans, she says. “That's the language that's out there.”
Deutschland Neuesten Nachrichten, Deutschland Schlagzeilen
Similar News:Sie können auch ähnliche Nachrichten wie diese lesen, die wir aus anderen Nachrichtenquellen gesammelt haben.
The Pentagon Inches Toward Letting AI Control WeaponsDrills involving swarms of drones raise questions about whether machines could outperform a human operator in complex scenarios.
Weiterlesen »
AI Might Help Us Decode Whale LanguageScientists are wielding algorithms in hopes of understanding how the mighty mammals communicate.
Weiterlesen »
DAOs: Where humans may fail, AI could succeedAre DAOs the next step forward towards a decentralized and non-hierarchical future? DotNever argues that once implemented, decentralized autonomous organizations could revolutionize the way we think about work.
Weiterlesen »
The Ethical Debate About Whether AI Ought To Warn You When The Self-Driving Car That You Are Riding In Is About To CrashQuick, you are inside a self-driving car and it is about to get embroiled in a car crash, what should you do? And what should the AI driving system do? Tough questions, for sure.
Weiterlesen »
This AI Predicts How Old Children Are. Can It Keep Them Safe?Yoti’s tech may be enticing for Big Tech companies: It works out if you’re under or over 13, the age most social media platforms require to create an account.
Weiterlesen »
iOS 15.4 makes Shortcut automations way less annoying to useA huge change for iOS power users
Weiterlesen »