[Tech Thoughts] Companies advocating for an AI future should be liable for its mistakes

The companies pushing for an AI-enhanced future should also be supporting their products instead of foisting the responsibility of fact-checking potentially harmful statements onto a not-so-AI-savvy public

A regional court in Germany found search engine operator Google liable for wrong answers made by its “AI Overviews” feature late in May.

Of course, Google has said it’ll challenge this ruling on false claims made by its AI Overviews, but this ruling by a German court could very well start the ball rolling on further actions seeking to tame or otherwise make AI companies accountable for errors made by their products.

Let’s break down what happened, and what it could mean if more people get on board a similar line of reasoning.

In Munich court, Google held liable for false claims on AI Overviews

The Decoder on June 9 released a story about a May 28 ruling, in which the Regional Court of Munich hit Google with a temporary injunction barring the company from posting false claims about two Munich publishers.

Google’s AI Overviews falsely connected the publishing companies to scams, subscription traps, and other shady practices in some search queries. According to the court, the AI mixed up information about other, genuinely sketchy companies with the plaintiffs. It made connections that were not in the linked sources, and Google did not respond appropriately to a cease-and-desist letter.

In a June 15 report by the Transparency Coalition, Google argued it “merely operates a search engine that automatically displays the data and information of third parties in accordance with search queries. The defendant is therefore not itself responsible for data processing, does not adopt the information of third parties in the ‘overview with AI’ as its own, and is accordingly only liable if it is made aware of an obvious infringement of rights.” The court did not find this argument persuasive.

While prior case law established that search engines were not legally liable for the results turned up in response to a query, the Regional Court of Munich held that “because the AI ​​used by the defendant independently compiles the information in the overview and summarizes it into a summary text, this constitutes an independent presentation for which [Google] is responsible.”

The court found Google liable “because the disputed display of the ‘results with AI’ is not a mere display of search results, but rather content attributable to it.” It added that, “This ‘overview with AI’ represents [Google’s] own statement made by its own AI offered to users.”

According to DW, which also reported on the matter, judges said the AI overview constituted “a self-contained statement with independently comprehensible content,” with which readers were given no indication of any unreliability in the content. The court told Google to stop spreading the false claims and to bear 80% of the legal costs.

In a statement to The Decoder, a Google spokesperson said they will challenge the ruling.

“We invest deeply in the quality of AI Overviews to ensure that the overwhelming majority of responses provide accurate information, and they are designed to reflect the information that exists on the web. We’re carefully reviewing this decision, which is not yet final,” the spokesperson said.

Human oversight needs a sufficiently human response

As we’ve tackled before. AI needs human oversight. However, it can’t just come from one angle.

While the needed human oversight of AI has to come from all sectors and stakeholders if you want it to work, the companies pushing for this future should also be supporting their AI products instead of foisting the responsibility of fact-checking potential harmful statements onto a not-so-AI-savvy public.

If some information is wrong or missing, humans should be at the helm of corrections and fact-checking these AI-made mistakes.

While I admit I do not like the idea of checking a robot, it may very well be good-paying work should Google and other companies opt to bankroll fact-checking and coherence teams that review claims of AI hallucinations.

The German publishers in this case saw something wrong, asked for a remedy to fix the reputation-hurting situation, and Google… apparently didn’t do anything in a timely manner to fix the mess it made.

Tech companies need to do better when it comes to creating AI monstrosities and leaving people to pick up the pieces after some egregious wrong has been committed by the AI monster. Hopefully, this ruling in Munich sticks, and maybe it’ll lead to further precedents later down the line. – Rappler.com