Google’s AI Overview & AI Safety

The impact of AI Overview on Websites

• Kevin Indig of Growth Memo conducted a study of 1,675 search queries in the health vertical and found that AI Overview appeared in queries 42% of the time and a cited domain saw an average traffic decrease of -8,9% when included in the AI Overview Results.

• AI Overview generates summaries at the top of search results with links to a few publications to which the information was sourced.

• Recently Google has come under criticism for the results produced by AI Overviews, with results from a “cheese not sticking to pizza” query returning results of adding 1/8 of non-toxic glue to the sauce to give it more tackiness (see screenshot below).

Why is this worth knowing?

Because of this change, you'll possibly need to review your content marketing plans, particularly for blogging. You should look at different ways to drive traffic to your blog posts - like social media, email newsletters, being a guest on podcasts, or paid ads. But, it's still really important to create high-quality, factual content. That way, your stuff is less likely to get misrepresented in search engine results and your organic content marketing will keep working well.

Discussions of AI Safety Summit

• During the AI Safety Summit, leading tech companies including Microsoft, Amazon and IBM have pledged to publish safety measures when developing foundation models.

• The companies have all agreed not to develop or deploy an AI model if the risks it poses cannot be controlled or mitigated.

Why is this worth knowing?

This is important for people using AI models because it shows that AI is being developed with a focus on safety. Users can feel confident that the AI models they use come from companies committed to minimising and controlling potential risks. This builds trust and confidence in AI technologies for users.

OpenAI Board Forms Safety and Security Committe

• OpenAI's Board has formed a Safety and Security Committee to make recommendations on critical safety and security decisions for all OpenAI projects within 90 days.

• Following the full Board’s review, OpenAI will publicly share an update on adopted recommendations in a manner consistent with safety and security.

Why is this worth knowing?

It shows they are seriously committed to developing AI responsibly by putting strict safety and security rules in place. This builds user trust. However, some people are critical because the new committee is made up of OpenAI's own executives. This means the evaluation of the safety measures won't be fully independent.

Espresso News Hits

• WhatsApp may add a feature to let users create AI-generated images for chats.

• Mastercard's use of generative AI doubled how many compromised cards it detects and sped up notifying customers by 300%.

• Microsoft has brought its AI assistant Copilot to the Telegram app.

• Vox Media and The Atlantic sign content deals with OpenAI

Post of the week