Google Search Generative Experience, or SGE, launched as an experimental search interface in May of 2023. Users gained access to the new search experience by joining a waitlist with their Google account through the new Search Labs program. Search Labs allows users to toggle experimental features on and off once they make it off the waitlist.
According to Google Search Labs the SGE experiments (generative AI in Search and SGE while browsing) will both end December 2023. With the end of the experiment approaching the question is – will Google SGE survive and become part of the default search experience?
Defaulting to SGE
After months of having the experience enabled it’s easy for some SEOs to forget SGE isn’t the default for everyone. But that’s likely to change soon. Many pundits and experts, including Barry Schwartz of Search Engine Roundtable, believe that Google will rollout SGE as part of their default search experience for all users in the US sometime in Q4 2023.
While most people are familiar with AI creeping into their everyday lives, a change this massive to one of the internets most used websites will be extraordinarily significant. So, if you haven’t seen it before, Google Search could look like the screenshot below for you and everyone else sometime before New Years.
But before making any assumptions we should consider what could prevent Google from defaulting to SGE by end of year.
Concerns & Challenges
Like many SEOs, I was optimistic and impressed by SGE at first. I completely agreed with Eric Enge’s initial review of SGE calling it a promising start and the overall user experience solid. After roughly three months of use I have a variety of concerns with SGE going live to the public and becoming the default for Google Search.
The three major challenges for Google SGE to overcome are:
- AI hallucinations
- Minimal utility over traditional search
- Cybersecurity policies and regulatory pressures
AI hallucinations are when a chatbot or other generative AI system claims a falsehood as a fact. They are often thought of as funny, but they can also be incredibly harmful and dangerous. And many question if the problem of AI hallucinations is even fixable.
Many SEOs testing SGE were happily surprised at first. For many the results felt sufficient and the novelty of the experience was exciting. Months later, the minimal utility over traditional search feels evident. Journalists and testers have complained about SGE being slow, verbose and not really a benefit over traditional search. The usefulness of SGE could face further strain if/when ads are introduced – as some have pointed out, the lack of ads has been one of the primary benefits of the experimental experience.
Cybersecurity policies and regulatory pressures are also likely to become a challenge for Google defaulting to SGE. Corporate and academic polices limiting the use of AI could severely limit the usage of SGE overall. And various branches of the government could apply regulatory pressure at any time – which seems increasingly likely due to other concerns being raised (hallucinations, privacy etc.).
Get your popcorn ready. With the December 2023 experiment deadline looming it’ll be interesting to see if SGE goes mainstream.
The AI Gold Rush
Even the most well-founded concerns regarding generative AI are unlikely to slow the gold rush. The desire to launch AI products from a marketing and branding perspective is likely to override any concerns or limitations of generative AI experiences. So yes, Google SGE is likely to survive 2023 and go live to the general public – but many SEOs are questioning if it should.
Like autonomous driving, AI-driven search experiences may be inevitable, but the risks of rushing these developments shouldn’t be overlooked for profit. Google SGE will be a huge gamble for the tech giant and at the risk of being overdramatic – it feels like the future of search is held in the balance.
There is a third option outside of defaulting to SGE or ending the experiment by year’s end – Google could extend the experiment into 2024. After extensive testing and research, it is my point of view that SGE would benefit from further refinement and some fundamental changes before becoming the default. And those improvements are less about the technology itself and more about having a dialog about ethics and incorporating those considerations into the technology.
If Nintendo spent a year testing and correcting bugs before launching Zelda Tears of the Kingdom, Google can surely invest more time to ensure quality and limit the risks of defaulting to SGE. Maybe that’s comparing apples to bananas. And maybe everything will be smoothed out by Christmas. But if I were in Sundar’s shoes, I’d definitely take door number three.
Update: less than 24 hours after publishing this article we saw reports of Google rolling out to a wider audience and displaying a an overlay promoting SGE.