400tmax
Over the past two decades, Google (NASDAQ:GOOG) (NASDAQ:GOOGL) has grown to create various revenue streams, but its core business continues to be advertising through its Search engines, most notably ‘Google Search’. Google’s Search advertising revenue has been slowing down recently, and fears are mounting over rising competitive alternatives. The tech giant has indeed been heavily investing in AI to keep up with innovation and stay relevant as a Search destination, though it could be innovating at the expense of its own future advertising revenue.
Google Search is Alphabet’s flagship service, offering an incredible advertising avenue for website publishers and digital retailers. Search activity incurs high-intent, enabling advertisers to target users exactly when they are actively seeking those specific products/services. In Q3 2022, 73% of Google Advertising revenue came from the ‘Google Search & other’ segment, but its year-over-year growth rate slowed to 4%. Executives cited the tough macroeconomic conditions as the main cause for the slowdown, though investors are becoming increasingly wary of new alternatives to the way people search.
ChatGPT – a threat to Google?
Google is currently the undisputed king of Search, though the rise of new platforms and services can indeed challenge Google’s moat as Search activity evolves. The latest buzz is around ‘ChatGPT’, a Natural Language Processing [NLP] software programmed to engage in text-based conversations with people in a natural, human-like manner.
OpenAI, the creator of ChatGPT, proclaims various capabilities of its latest NLP software innovation.
ChatGPT homepage
In fact, within 5 days of launch in late November 2022, ChatGPT reached 1 million users, as people experimented with the software to ask natural questions, write essays, and even generate step-by-step instructions for completing complicated tasks. Despite the excitement around its capabilities, it is important to make clear that ChatGPT has no corporate applicability at the moment.
In fact, OpenAI CEO Sam Altman has cautioned on ChatGPT’s limitations, making it clear that not all answers produced by ChatGPT may be factually correct, undermining its business use case presently. “The way it works currently is people use ChatGPT and then go to Google to verify the results,” according to AI consultant Tobias Zwingmann. From this perspective, claims that ChatGPT threatens Google’s Search dominance are ludicrous.
That being said, ChatGPT is meant to be an advancement of GPT-3, which is available through Microsoft’s (MSFT) cloud service Azure for corporate use, and offers more reliable answers. Note that Microsoft is also an investor in OpenAI. Hence, Google will certainly need to keep an eye on advancements in the reliability and use case of ChatGPT, and how OpenAI and Microsoft incorporate it into products and services, given that Microsoft competes with Google in both the cloud industry, as well as web Search (through Bing). Google’s executives and shareholders cannot underestimate the competitive potency of rising services, no matter how negligible they may seem, as competition can indeed manifest itself in unexpected manners. For example, TikTok primarily became popular as a source of entertainment, posing a threat to YouTube. However, nowadays the younger generation is also turning to TikTok for other reasons, such as financial advice from so-called “experts”, thereby also threatening Google’s Search business.
The point is, Google will need to prove to investors that it is able to effectively innovate and continue offering useful and relevant Search solutions. The introduction of ChatGPT does raise questions about Google’s own AI initiatives, and whether the tech giant will be able to competently fend off rising competition to maintain its Search dominance.
Google’s LaMDA
Google’s LaMDA is the company’s own AI-powered NLP software, engaging with users in a conversational manner. The company has been working on LaMDA for years, raising the question of why Google has not officially rolled it out yet, and whether the tech giant is moving too slowly in terms of AI innovations.
At a recent all-hands meeting at Google, Jeff Dean (Head of Google’s AI division) highlighted Google’s own AI prowess and said that “we are absolutely looking to get these things out into real products and into things that are more prominently featuring the language model”. Moreover, CEO Sundar Pichai proclaimed that he sees next year as a “point of inflection” for how AI is applied in Search.
Responding to employees’ concerns regarding Google being perceived as lagging on AI innovation amid the ChatGPT-hype, both Google’s CEO and AI head highlighted that LaMDA has similar functionalities, “but that the cost if something goes wrong would be greater because people have to trust the answers they get from Google”. Indeed, the rollout of ChatGPT revealed flaws that manifested in the form of answers including non-factual information, as well as producing sexist lyrics when prompted to write song lyrics.
Given Google’s size and position, the stakes are much higher for the tech giant, hence the company can’t afford to rollout AI software that is still work-in-progress. If people feel they can’t trust answers/ search results from Google, it could raise the risk of users seeking answers from elsewhere, such as Bing or professionals on TikTok. Therefore, Google is wise to prioritize reliability, rather than a speedy rollout, in the interest of maintaining its dominance as a web Search engine.
The head of AI further explained that “you can imagine for Search-like applications, the factuality issues are really important and for other applications, bias and toxicity and safety issues are also paramount”. The last thing Google needs is reputational damage due to the spread of misinformation on its platform, a plight that advertising rival Meta (META) continues to suffer from. Hence, Google’s executives are taking the right approach by delaying public rollout until they can be confident that it will not cause wide-scale harm among users. Trust and optimal user experience are key to maintain user engagement.
Furthermore, searching for answers on Google is a habit that is well-embedded into people’s behavior, and displacing such behavioral habits poses an uphill battle for rising competitors. This buys Google time to work on and perfect its own AI advancements, and the company is using this to its advantage, by only rolling out LaMDA to the general public when the risk of embarrassing failures is adequately subdued.
No AI-powered solution will be 100% perfect upon launch, and such technologies generally go through various iterations to enhance its accuracy. As the number of people using the technology increases, the software is able to learn more and more, improving its ability to offer more useful answers overtime. On that note, Google’s sheer size as a mega-cap employing thousands of employees is also a leverage that it possesses over rising competitors. Moreover, given Google’s large employee base, it has the luxury of testing out its own products internally before rolling it out publicly, which it is already doing, thereby improving the chances of any prominent/ obvious bugs being resolved internally, subduing reputational risks upon public rollouts. Competitors like OpenAI are unable to test products internally on the scale that Google can. The public rollout of ChatGPT as a ‘work-in-progress’ indeed entails reputational risks, as it risks leaving a negative impression among users.
Note that in August 2022, Google also released LaMDA on its AI Test Kitchen App, allowing certain U.S. users on Android devices to experiment with the software, give feedback on the relevance/ usefulness of the answers produced by LaMDA, as well as allow Google to learn how people might use the software going forward.
Google LaMDA versus advertising efficacy
Google’s LaMDA indeed reflects the tech giant’s AI prowess, and its ability to stay innovative in the face of rising competition. Though the deployment of new technology also creates new challenges in terms of monetization. Google’s bread and butter comes from Search advertising revenue, enabling advertisers to place ads right at the top of Google Search results to target potential consumers in an opportunistic manner. Google’s ambitions to enhance Search with the use of LaMDA has its share of opportunities and risks.
LaMDA’s ability to engage in natural conversations enhances its ability to understand what people need in a more coherent and contextualized manner, as opposed to generic Search queries which can yield a broad array of Search results and advertisements, a certain amount of which may not be useful to the user. Everyone has entered a Google Search query which generates irrelevant ads at the top of Search results that one simply scrolls past. Contrarily, a user’s ability to ask LaMDA questions in a more natural manner, and the software’s ability to carry on conversations to learn more about the user’s query type, enhances Google’s ability to grasp user intent, thereby enhancing its targeted advertising potential.
Offering people more relevant answers/ search results enhances Google’s ability to keep users engaged, thereby subduing the risk of people searching for solutions on alternative platforms like TikTok. Engaged users are also more likely to come to Google for commercial searches, enabling more business activity to occur through the platform, and thereby improving the efficacy of Google Search ads.
On the other hand, how Google deploys LaMDA into Search will be crucial in determining how well the tech giant will be able to maintain or improve advertising potential. LaMDA’s enhanced capabilities are intended to deliver answers to users more directly and intuitively. For example, someone querying about a health issue would supposedly be given answers and recommendations by LaMDA, which scrapes the web for the most relevant and useful answer. Google’s apparent focus on ensuring the answers provided are truthful would indeed augment the appeal of using LaMDA, and in turn reduces users’ needs to visit various other websites relating to their query to find the answer/ content they are looking for. A potential decline in people visiting websites would consequently undermine the appeal of Google Search ads, as the allure of appearing on top of Google Search results fades away if LaMDA provides the answers people are looking for directly.
Furthermore, because LaMDA undermines publishers’ discovery potential on the web, it would also hurt Google’s advertising revenue through the Google Network segment, which comprised around 15% of Google’s advertising revenue in the first three quarters of 2022. More specifically, AdSense (part of Google Network), which enables publishers to sell ad spaces on their websites to advertisers, would witness a notable decline in ad revenue if LaMDA hinders people’s needs to visit websites. Google is indeed walking a fine line between perfecting the usefulness/truthfulness of answers provided by LaMDA and enabling publishers to continue witnessing adequate web traffic/ growth in web traffic. Publishers would certainly not be happy to witness a decline in web visitors as a result of LaMDA gaining prominence off of their own content. Keep in mind that certain publishers are already suffering from Google’s unethical business practices relating to its Search service, resulting in various antitrust lawsuits. Potentially losing out on web traffic due to LaMDA may only aggravate their experience of listing on Google, and could encourage them to explore other avenues. Hence, Google may have to completely re-imagine how it enables publishers to monetize their websites, and how the tech giant generates advertising revenue itself, if it indeed incorporates LaMDA into its Search functionalities going forward.
Aside from Search queries where web surfers seek to learn about something, Google Search is also used to look for products and services. Google has been striving to encourage users to start their product searches on Google rather than Amazon (AMZN), in the interest of inducing high-intent shopping activity through its own platform, and thereby promoting to efficacy of its Search ad solutions. Google users will be familiar with the fact that when they search for a particular product to buy, relevant product listings from various digital retailers appear on top of the Search results in the form of image ads. With the current advertising model, the highest-bidders appear in Search results when certain keywords are triggered.
Though the introduction of LaMDA in Search functionalities could prove to be a game-changer. LaMDA’s conversational-style and its goal to deliver the most useful answers/ solutions could inevitably lead to the software recommending products/ services to its users, depending on how personal it is programmed to be. Moreover, it poses a new dilemma for Google, as it will need to decide whether to program LaMDA to suggest the best-suited products/ services based on what it learns about the user, or to program it so that it naturally recommends the products/ services of advertisers with the highest bids. Google essentially faces a tug of war between optimizing the user experience and upholding the efficacy of its advertising business.
While its ability to better understand the user’s need through a human-like conversation could indeed improve targeted advertising potential, Google’s LaMDA recommending products/ services could create new problems, as it could be on the hook if LaMDA ends up recommending a brand that users end up having a negative experience with. For instance, a crypto firm could be the highest bidder for keywords related to financial services and investing. If the company turns out to be a scam, users may become averse to using LaMDA again, or worse, Google’s services altogether. Conversely, with the current model, Google has a more detached approach to advertising, whereby they are simply displaying product/ website listings that are relevant to the user’s Search query, and not recommending any product/ service based on striving to understand what the user is after through a human-like conversation. Google will need to strike a balance between ‘listing’ and ‘recommending’ through LaMDA, and will likely need to program it so that it offers rather diplomatic answers to avoid being on the hook when users end up having negative experiences with certain products/ services.
Summary
Despite the recent hype around ChatGPT, it does not pose a threat to Google’s Search dominance, though it does raise questions about Google’s own AI prowess, and whether the tech giant will be able to competently fend off rising competition going forward. Google has indeed been working on its own NLP software ‘LaMDA’, enabling engagement with users through human-like conversations. The AI-powered solution could enhance Google’s targeted advertising potential, but becoming too personal creates new dilemmas. Delivering increasingly accurate answers through LaMDA in response to Search queries may eliminate users’ needs to visit websites, disheartening publishers and undermining the appeal of Google Search ads and Google Network partners. The evolution of AI forces Google to choose between optimizing the user experience and upholding the efficacy of its advertising business.
Google has become the tech giant that it is today thanks to the advertising potency of its Search engines. Though evolving search behavior and advancements in AI are creating new Search solutions that will inevitably transform Google’s core business model and ways of monetization. How exactly Google goes about the transformation can only be determined over time.
Alphabet consists of various business divisions, including Google Cloud and Pixel hardware devices, which investors should take into consideration when making investment decisions. Given that this article solely focuses on Google Search, a neutral ‘hold’ rating will be assigned to the stock.