Contents
Google Trials AI-Generated UK Search Results
‘Language models are the future of search engines’ – if you are abreast of the ever-growing conversations surrounding artificial intelligence you will have undoubtedly encountered this idea, along with the arguments to the contrary.
Many raise issues with models such as Chat-GPT being used as search engines, citing their use of data that isn’t live, and a tendency for AI to piece together what it thinks is the right answer without any robust measures to ensure complete accuracy. Now, Google has stepped into the ring, setting out to trial selective AI-generated search results – dubbed “The Search Generative Experience” – within the UK.
SEO marketers, web developers, copywriters – is this the beginning of the end? Is the looming shadow of AI set to overtake the need for websites for information? Will Google become simply an interface for AI, websites relegated below the fold to collect dust in case a user actually wants to do it the old-fashioned way and look for a result themselves? We wouldn’t be so hasty.
Google can always be trusted to be ahead of the game, but equally cautious – these results are only being tested as an ‘overview’ for specific results. This likely will not include complex or controversial topics, where AI blunders are rife, and instead will deal with more benign topics/queries with the intention of “maintaining information quality”. “How to get marks off painted walls” is one Google says people find helpful – demonstrating the more restricted capabilities of their new feature. The last thing Google wants is a slew of screenshots floating around social media of their model giving people wildly incorrect information.
Within the US, consumers have reportedly found the feature highly useful. Google itself reports that its aim to ensure traffic still goes to sites has been successful, with the model even “showing more links to a wider range of sources.” It’s important to remember how much Google has invested into the monolith that is Google search and Ads, relying on their 50 billion indexed sites as much as those sites rely on them.
Ever since AI exploded into the common conversation, it has been only a matter of time before Google adopted it in such a fashion. As it stands, however, it seems their caution in favour of accuracy and safety along with the current state of technology means that this will not overtake traditional search anytime soon.
Meta Unveils New System to Identify AI-Generated Content
Meta recently announced an update to its AI labelling policy, broadening its definition of “manipulated media” to extend beyond AI-generated videos, now encompassing deceptive audio and images on Facebook, Instagram, and Threads.
A significant aspect of the new policy is its sensitivity towards being perceived as restrictive of freedom of expression. Instead of opting for content removal, Meta now chooses to label problematic content. Meta has introduced two labels, “Made with AI” and “Imagined with AI,” to clarify content created or altered with AI.
The AI-generated content will be identified based on AI authorship signals and self-reporting:
“Our ‘Made with AI’ labels on AI-generated video, audio, and images will rely on our detection of industry-shared signals of AI images or people self-disclosing that they’re uploading AI-generated content.”
Content that is notably misleading may receive more prominent labels to enhance user understanding. Harmful content violating Community Standards, such as content inciting violence, election interference, bullying, or harassment, will be subject to removal, irrespective of whether it is human or AI-generated.
Meta’s original AI labelling policy was crafted in 2020, primarily targeting deceptive videos. However, with technological advancements, Meta’s Oversight Board recognised the necessity for a revised policy, extending to AI-generated audio and images.
Meta’s policy update was informed by extensive feedback from stakeholders and the general public. The company engaged with over 120 stakeholders in 34 countries and conducted public opinion research with more than 23,000 respondents across 13 countries. A large majority (82%) favoured warning labels for AI-generated content depicting individuals saying things they did not say.
Meta’s revised policy underscores the importance of transparency and context for AI-generated content. Content removal will be based on violations of Community Standards, with labelling being the preferred response for potentially problematic content.
Google Suggests Over-focussing on Links Could be Useless
Google recently suggested that “over-focussing” on links could be a futile endeavour, contributing little to improving one’s website.
In response to a question regarding backlinks, Google stated that fixating excessively on links might not yield substantial benefits. This assertion aligns with Google’s recent trend of de-emphasising the significance of links over the past half-year.
Google further clarified that backlink checkers do not select specific sites and their backlinks but instead crawl the web to map link relationships between sites. However, due to the vastness of the web, these tools prioritise what they crawl, resulting in variations among different tools.
Regarding the subjective nature of counting links, Google highlighted that there is no universally accepted method, as each tool collects data differently. Therefore, discrepancies in reported link counts are inevitable.
They emphasised that website owners should not overly fixate on the sheer number of links, as search engines can discover websites through various means, such as sitemaps. He hinted that focusing too much on links could lead to misallocation of efforts that do not contribute significantly to overall website improvement.
This sentiment echoes Google’s recent stance, indicating a diminishing emphasis on the importance of links in ranking algorithms. Over the past six months, Google has subtly conveyed this shift, evidenced by alterations in their documentation, including the removal of the term “important” in reference to links as a ranking factor.
In conclusion, recent statements and actions suggest a reevaluation of the traditional emphasis on links in website ranking algorithms, advocating for a more holistic approach to website optimisation.
