The advent of AI Overviews, now known as SGE, has sparked a new era of search dominated by artificial intelligence. This transition began when users started interacting with ChatGPT, OpenAI’s chatbot, in November 2022. However, the integration of AI into search engines could have been smoother.
During Google’s I/O event in 2023, the company officially announced the beta testing of AI Overviews, touting it as the future of search. Despite the grand announcement, Google has taken a conservative approach to the rollout, limiting access to logged-in users in the US. This cautious approach may have been prompted by the criticism AI Overviews faced from the media and users. Additionally, the computational resources required for AI-powered answers and the monetization of these answers are still under development. Thus, it seems that the official launch of AI Overviews is more like a large public beta test for Google to gather more data and refine its language learning models (LLMs).
Google appears to have implemented internal mechanisms and criteria to assess the quality of the AI Overviews-generated answers. These criteria have become stricter post-launch, with Gemini, Google’s AI model, generating answers in niches where it is more confident in its responses. SE Ranking’s research indicates that AI Overviews are most commonly triggered in the Relationships niche (26.62%), followed by Food and Beverage (24.78%) and Technology (18.11%). Conversely, niches related to user well-being such as Healthcare (0.44%), Legal (0.34%), and News and Politics (0.24%) trigger AI Overviews less than 1% of the time. This cautious approach in sensitive niches is likely an attempt to mitigate potential harmful effects caused by AI hallucinations and biases.
However, even in niches where Google frequently triggers AI Overviews, there have been instances of harmful answers. For example, an AI Overviews response suggested adding glue to pizza sauce, citing a joke post on Reddit. Integrations with platforms like Reddit, Quora, Stack Overflow, and Wikipedia provide a vast amount of human-generated data, but the moderation of such content was not designed to train LLM models. Platforms like Reddit are known for their abundance of jokes and inappropriate memes. Interestingly, while OpenAI has also collaborated with Reddit, its chatbot, ChatGPT, does not produce potentially harmful answers as frequently as AI Overviews. This discrepancy indicates that Gemini may rely more on directly citing Reddit without in-depth content analysis. OpenAI’s focus on creating an LLM with improved “critical thinking” capabilities has allowed it to outpace Google in the AI race. Although Google had DeepMind at one point, few people are aware or recall this fact.
Nevertheless, Google still has a significant advantage in terms of data volume. With billions of daily searches and access to data from 3 billion Android users, Google’s data pool for training its models is unparalleled. Its integration into the vast Android system is an exclusive advantage that no other LLM model can achieve. While OpenAI has integrated its LLM into Apple’s OSes, the collaboration is non-exclusive, as Apple adheres to a no-vendor-lock policy. Additionally, Apple has developed its own LLM and plans to integrate other models like Claude.
However, given OpenAI’s partnerships with Reddit, Apple, and Microsoft, the organization has the ambition and resources to rival Google. It is entirely possible that OpenAI may launch its own answer engine in the future. Compared to Google, OpenAI is more agile and tends to make bold and emotionally driven decisions. These decisions include the departure and subsequent return of Sam Altman and the recent controversy involving Scarlett Johansson’s voice during the launch of ChatGPT-4o, among others. However, OpenAI is treading carefully when it comes to launching an answer engine that directly competes with Google. The organization is weighing the pros and cons, primarily concerned with addressing AI hallucinations, biases, and the potential spread of misinformation. OpenAI may choose to leave the launch of answer engines to other players who plan to utilize OpenAI’s LLM instead.
One of the main challenges AI Overviews faces is the abundance of data available for training LLMs. According to SE Ranking’s research, scrolling through organic search results is increasingly difficult, with featured snippets appearing next to AI Overviews 45.39% of the time. When both featured snippets and AI Overviews are present, their sources match 61.79% of the time, resulting in duplicate information. Furthermore, ads now accompany AI Overviews 87% of the time, up from 73% before the official launch during Google I/O 2024. This co-occurrence of AI Overviews, featured snippets, and ad blocks may be a step towards training the model to compile an AI Organized Search Results Page. However, the placement of genuinely helpful answers further down the page and the presence of AI hallucinations and duplicated information could alienate users. This could lead to a decrease in user loyalty and trust in Google, potentially driving them to adopt alternative search methods such as ChatGPT. OpenAI already offers more than just search answers with Gemini’s integration into G-Suite, although its success remains uncertain. In the worst-case scenario, some users may blindly trust AI Overviews’ answers without fact-checking. It is the responsibility of Google and other LLM models to address these challenges seriously and ensure the accuracy of their answers.
The shift to AI Overviews and an AI Organized Search Results Page, prioritizing quick and accurate answers, could pose challenges for the SEO industry, small and medium-sized businesses, news outlets, and independent creators. As users find their queries satisfied by AI, they may no longer feel the need to scroll through organic search results or delve deeper into specific sources. This change in user behavior could make it challenging to drive traffic and reach target audiences organically. Consequently, more businesses, media outlets, and creators may migrate to social media, email newsletters, or niche platforms, and some may even face financial struggles and potential bankruptcy. Maintaining a balance between transitioning to an answer engine and supporting business users will be essential for Google’s success. The possibility of preserving the traditional SERPs as a feature called “Google Classic” remains uncertain. If not, we may witness a more fragmented and less diverse internet landscape.
Determining who will lead the GenAI era of search is a matter of weighing the advantages of both Google and OpenAI. Google’s massive user base and exclusive integrations are unmatched, while OpenAI’s focused development efforts and influential partnerships position it as a strong competitor. The ultimate winner of this race will likely be determined by who can best balance technological innovation with user needs and ethical considerations.
Overcoming challenges such as AI hallucinations, addressing user adoption and trust concerns, and supporting businesses and media will be crucial in leading the search revolution. The need for fact-checking remains a primary challenge for all AI models. ChatGPT often declines to answer certain questions, reminding users that it is just an AI model, while Google reduces the number of AI Overviews-triggered answers to avoid potential inaccuracies.
Despite some setbacks, Google still stands a chance to emerge victoriously in the GenAI search race. Integrating Gemini into the Android system provides an opportunity for 3 billion people to interact with AI in their daily lives. For instance, Android users could use AI to determine the freshness of a tomato simply by taking a photo and speaking. If a significant portion of the global population adopts Gemini, it would become one of the most widely trained AI models in the world. However, Google needs to find a way to acquire the necessary resources without consuming electricity at the rate of a large European country.
It is essential to keep in mind that OpenAI’s unpredictable nature and the rapidly changing landscape of AI could completely reshape the search industry. Current advancements and future developments that we are currently unaware of may significantly impact the outcomes of this race.
Source link
