Although science fiction films from the 1980s and '90s cautioned us regarding the possible destruction of civilization through artificial intelligence, the actual outcome up until now has been considerably more subdued.
Skynet was intended to oversee the emergence of destructive robots known as Terminators, which could only be halted through time travel and convenient narrative inconsistencies.
The artificial intelligence from "The Matrix" films similarly fought against its human makers, imprisoning most of humanity within a simulated reality and forcing the resistance movement into hiding.
Related: Meta invests massive funds to outpace Google and Microsoft in crucial competition
Fairly speaking, the artificial intelligence systems from OpenAI, Google Gemini, Microsoft Copilot, and various others threatens to annihilate humanity, but not consistently And it seems that the technology poses little threat to our likelihood of surviving.
However, that does not imply this revolutionary technology isn't generating other significant issues.
The primary concern people have about AI today revolves around how companies develop their models.
Large language models such as OpenAI's ChatGPT require substantial data intake to pass the Voight-Kampff test from "Blade Runner," with much of this data being under copyright protection.
At present, the potential for AI lies within the jurisdiction of the courts rather than with software engineers.
This week, the courts issued a landmark decision with potential far-reaching consequences.
Judge delivers verdict in AI case, leaving numerous queries unanswered
This week, Judge William Alsup from the U.S. District Court for the Northern District of California decided that AI firm Anthropic and others are permitted to use published books for training their AI systems without obtaining permission from the authors.
This decision might establish a key legal example for the many other pending AI copyright cases.
More on AI:
- Gemini, ChatGPT might fall behind in the AI competition against well-funded competitors.
- Anthropic highlights AI limitations through terminating blog experiment
- The recent improvement of Amazon's Alexa AI has turned out to be even more disappointing than anticipated.
Three authors sued Anthropics for allegedly violating copyright laws by using millions of pirated books to train their large language model, however, Judge Alsup ruled in favor of Anthropic.
The copies utilized to educate particular LLMs were deemed a fair use," Judge Alsup, who has previously overseen landmark cases such as Oracle America v. Google Inc., stated in his decision. "All factors except for the character of the copyrighted material support this conclusion. This technology represents one of the most revolutionary developments we may witness throughout our careers.
The Anthropics ruling has mixed implications for the fair use of copyrighted content.
Ed Newton Rex, the CEO of Fairly Trained—a nonprofit organization that pushes for ethical compensation for those whose work is used to train data-driven large language models—offered an insightful perspective on the ruling. Many news outlets portrayed it as a significant victory for AI corporations, but he saw things differently.
"The decision handed down today in the lawsuit between the authors and Anthropic regarding copyrights presents a nuanced outcome. It doesn’t represent the victory for AI firms that certain news headlines imply—there are both positive and negative aspects to consider," he stated. A long Reddit post this week .
To put it briefly, the judge stated that Anthropics' utilization of copyrighted books constituted an infringement, yet their learning from non-copyrighted material was deemed fair use.
Therefore, Anthropics faces liability for using the content without authorization, yet the judge decided that they do not require the author’s consent to develop their models.
This indicates that Anthropic's fair use claim was upheld in court; however, the decision might not have as broad an impact as it appears.
"This should not be considered a universal decision stating that all generative AI training qualifies as fair use. Different circumstances could lead to varying outcomes," stated Newton Rex.
The Copyright Office has previously noted that certain AI models are more transformative compared to others — specifically, they highlighted that AI music models are considered less transformative. Lobby groups might claim that this ruling supports the idea that using generative AI for training purposes falls under fair use; however, that assertion is incorrect.
Related: Developers at Amazon have an unexpected rationale for their dislike of Generative AI