Google is set to introduce AI-generated ads in the near future

Advertisers have the opportunity to optimize their ad campaigns and potentially increase click-through rates by utilizing tools inspired by ChatGPT to generate text and images. Meta (formerly Facebook) is currently conducting tests on similar technology, recognizing its potential benefits for advertisers.

Over the past few weeks, Google has been actively promoting its generative AI tools, which can summarize search results, assist in drafting essays, and enhance images. Now, Google is showcasing how these tools can benefit its core business of selling ads.

The newly introduced generative AI systems for advertising clients enable on-the-fly composition of text that aligns with a user's search query. Additionally, these systems can generate product images, saving time and money on design work. These features expand the growing range of AI-based text and image generators that have been integrated into online services in recent months, following the success of ChatGPT and DALL-E, which sparked global interest in generative AI. Google, being the world's leading seller of online ads, has utilized AI programs for years to assist clients in targeting users and designing ads, including automated image resizing. With more powerful AI models capable of generating photo-realistic images, Google aims to demonstrate that its ad business, responsible for 80 percent of its total sales, can be even more appealing to advertisers.

The flurry of AI-related announcements by Google has bolstered shares of its parent company, Alphabet, indicating a decrease in concerns about the potential impact of ChatGPT-style web search on Google's search and ad businesses.

While Google is offering these new features to advertisers for free, they have the potential to increase revenue by encouraging businesses to place more ads or generate more clicks from consumers. Given Google's dominant position in online ad sales, the advertising industry is likely to be one of the first to widely adopt generative AI in their workflows. Jerry Dischler, the vice president overseeing Google Ads, states that these features enable the delivery of more relevant and visually appealing ads to users, offer greater creative freedom for advertisers, and enhance overall performance. However, Dischler refrained from discussing specific financial prospects for generative AI in ads.

As users who have experimented with AI chatbots or image generators may know, the output can be unpredictable or even objectionable, raising concerns about potential copyright infringement during their development.

To address these concerns, Dischler assures that Google will be diligent in monitoring the quality of images and text generated by the new features. Some of these features are already available to advertisers in beta form. Google is launching them more broadly than its top competitor, Meta, which recently announced an initial invitation to select advertisers for testing its own generative AI features.

Offering generative AI in ads is likely to be expensive due to the high computing costs associated with operating text and image generation models. At a recent conference, Meta AI executive Aparna Ramani stated that generating outputs from such models is 1,000 times more expensive than using AI for content recommendations and curating users' News Feeds.

One of Google's new features adapts the text of English-language search ads based on the user's search query and Google's data on the advertiser. Previously, algorithms would have to select text from a pre-written collection provided by the advertiser each time a search was performed. With the text generation option, a search for "skin care for dry sensitive skin" could trigger an ad for skin cream with the auto-generated text "Soothe your dry, sensitive skin." While this may not seem groundbreaking, aligning ads more closely with searches could increase the likelihood of user clicks.

Google also employs its text generation technology to offer a chatbot that provides ad customers with suggestions for search keywords and corresponding ad text. This feature, called "Help Me Create My Ad," leverages Google's data on past campaigns and analysis of the customer's website. Initially available to select US advertisers in English starting in July, this feature resembles Google's "Help Me Write" feature being tested in Gmail and Google Docs, which has been found to be a helpful creative aid but occasionally perpet