Zuckerberg hopes to use an AI system to fight fake accounts, fake news, and abuse of user privacy. He may have overlooked that the content posted by Cambridge Analytica was distributed to users through Facebook's algorithm.
On April 10-11, on Capitol Hill, Facebook CEO Mr. Zuckerberg faced two rounds of battles.
Prior to this, Cambridge Analytica collected personal data of more than 78 million Facebook users without their consent and used it to build models to analyze users’ political preferences. The models were used to deliver precise political content to voters during the 2016 US election. For this reason, the Senate and the House of Representatives launched a hearing against Mark Zuckerberg.
In these two hearings, Zuckerberg had to repeatedly explain Facebook's privacy policy, monopoly position, and the consequences of the "Cambridge Analytica" incident. The frequency of mentioning artificial intelligence (AI) was eye-catching.
Hearing or Facebook’s AI strategy conference?
In his discussion, Facebook has used artificial intelligence in multiple scenarios of content and information flow, such as:
"We have developed new artificial intelligence tools that can identify fake accounts that attempt to interfere in elections or spread false information. In three elections, we have promptly removed tens of thousands of fake accounts to prevent them from causing further interference (in the election). "
“If someone on the platform finds that there is content that violates the community’s terms, they will report it to us, and then we will review it. Now, we are developing and using more and more artificial intelligence tools to proactively identify harmful user behavior and flag it to our team.”
“As we sit here, 99% of content on Facebook related to ISIS and al-Qaeda is being flagged and removed by AI systems before anyone sees it. AI is proactively keeping the platform safe.”
"Identifying whether foreign actors are trying to influence elections has become one of our top priorities, and we will prevent it from happening again. We have done a lot of work, especially using artificial intelligence technology."
If the senators hadn't frequently asked sharp questions or made hilarious jokes, this would have been Facebook's artificial intelligence product launch conference. Not to mention that Facebook's stock price rose by nearly 8% during the two hearings.
Distrust of Facebook and AI technology
However, lawmakers are still more concerned about Facebook's data privacy policy than artificial intelligence. Obviously, if Facebook's AI really worked, Zuckerberg wouldn't have to sit here and face the hearing today. Two months ago, when the US Congress questioned fake news, Facebook still insisted that its user data had never been misappropriated. Where was AI at that time?
This involves a very important question: what exactly is AI?
Artificial general intelligence has been abbreviated to AGI to distinguish it from current AI components. It seems that any software system that uses neural networks, classical machine learning algorithms, or even text search can be called AI.
Don’t forget that after Cambridge Analytica studied the voter data, it used Facebook’s information flow algorithm and ad recommendation algorithm to deliver its own political content and native ads to the corresponding voters.
When some congressmen raised similar views or expressed a preference for stricter regulation of Facebook, Zuckerberg reluctantly said that they would hire 20,000 information reviewers to conduct self-censorship by the end of this year. Chinese netizens on the other side of the ocean said that this statement seemed familiar.
Why do we need 20,000 censors when we have artificial intelligence? Zuckerberg said,“If we want to fully rely on AI tools to review content, it will probably take more than five years.”
What? Five years from now, voters' opinions will be completely handled by artificial intelligence and algorithms? It seems that Mr. Zuckerberg of Facebook still doesn't understand what "information cocoon" means, and events like "Cambridge Analytica" will still happen again in the future.
Let's wait and see.
Historical articles (click on the image to read)
"Why is October 24th Programmers' Day?"
"This paper is poisonous!"
"How to explain artificial intelligence to relatives and friends?"
