[ad_1]
“I agree in principle that we should take responsibility, but I don’t think we have found the right set of terms to describe the process we are focusing on,” said Jonathan Stray, a visiting scholar at the Berkeley Center for Humanity. -Research the compatible AI of the recommendation algorithm. “What is amplification, what is enhancement, what is personalization, and what is recommendation?”
For example, the New Jersey Democrat Frank Pallone’s Anti-Malicious Algorithmic Justice Act will revoke immunity when the platform “knows or should have known” that it is making “personalized recommendations” to users. But what is personalization? According to the bill, it uses “person-specific information” to enhance the distinctiveness of certain materials relative to others. This is not a bad definition. However, on the surface, it seems to say that any platform that does not show the exact same content to everyone will lose the protection of Article 230. Even if you show someone a post from a person they follow, it can be said to rely on that person’s specific information.
Malinowski Act Protect Americans from dangerous algorithmsIf the platform “uses algorithms, models or other calculation processes to rank, rank, promote, recommend, amplify or similarly change the delivered or displayed information.” However, it contains algorithms that are “obvious, understandable and transparent to reasonable users” Exceptions, and lists some examples that meet the requirements, including feeds in reverse chronological order and ranking by popularity or user reviews.
This makes a lot of sense. One problem with engagement-based algorithms is their opacity: users hardly understand how their personal data is used to target content predicted by the platform to them. But Stray pointed out that it is not easy to distinguish a good algorithm from a bad algorithm. For example, ranking based on user reviews or voting yes/no votes is inherently bad. You don’t want posts with a single vote or five-star rating to be at the top of the list. Stray explained that the standard way to solve this problem is to calculate the statistical margin of error for a given content and rank it according to the bottom of the distribution. That technique-Stray took a few minutes to explain to me-is it obvious and transparent? How about something as basic as a spam filter?
“I don’t know if the intention to exclude systems that are sufficiently’simple’ would actually exclude any practical systems,” Stray said. “My suspicion is that it may not be.”
In other words, a bill to abolish section 230 of the algorithm recommendation exemption may ultimately look the same as the direct repeal, at least as far as social media platforms are concerned. Jeff Kosseff, author of the authoritative book on Article 230, Twenty-six words that created the Internet, Pointed out that even if Internet companies do not have legal protection, there are many legal defenses they can rely on, including the First Amendment. If the regulations are flooded with enough exceptions and exceptions, these companies may decide that there is an easier way to defend themselves in court.
[ad_2]
Source link
