The tech world is intrigued by a recently leaked memo by one of Google’s Senior Software Engineers, Luke Sernau.
In the memo, Sernau admitted that Google is fighting a losing battle in the AI space. Surprisingly, however, this admission did not pertain to their big tech rival, OpenAI.
“While we’ve been squabbling, a third faction has been quietly eating our lunch… I’m talking, of course, about open source,” said Sernau.
Open-source artificial intelligence is a type of AI technology shared freely and openly with the public, allowing anyone to access and use it. This open collaboration encourages developers to collaborate and share their insights to create innovative solutions.
In contrast, both Google and Microsoft-backed OpenAI own their respective AI technologies. Both these technologies are closed-source and proprietary, meaning they have complete control over their AI technology, who can access it and what they can do with it.
Sernau explains that this happened after Meta’s large language model, LLaMA, got leaked online, resulting in several coders and software engineers tinkering with the large language model. This solved a problem many big tech companies face with their AI tech: scaling and quickly implementing global iterations to their ******.
“The barrier to entry for training and experimentation has dropped from the total output of a major research organization to one person, an evening and a beefy laptop,” Sernau said in the leaked Google memo. “In terms of engineer-hours, the pace of improvement from these ****** vastly outstrips what we can do with our largest variants, and the best are already largely indistinguishable from ChatGPT.”
Currently, their large language ****** with 540 billion parameters have become a liability since “data quality scales better than data size.” The open-source community has been training on small, high-quality data sets that allow for better fine-tuning and eliminating the need to start from scratch when training their AI.
To solve this growing problem, Sernau pointed to Google’s past strategic decisions in two of their most successful products today: Android and Chrome. Both these platforms are released on an open-source basis, helping Google establish itself as a thought leader in these spaces. Sernau suggests Google do the same with their AI ******, even if it “means taking some uncomfortable steps, like publishing the model weights for small ULM variants.”
Whether or not open-source (and not OpenAI) is Google’s biggest rival is debatable. However, this leaked Google memo offers a new perspective on how we should expect the AI space to evolve. Would this mean a pivot towards open-source domination for Google? Or would they continue developing their existing AI products in a closed system?
Disclaimer: This leaked Google memo is only the opinion of one of Google’s Senior Engineers and does not reflect the company’s official stance.
More SEO News You Can Use
Google Search To Become “More Visual, Snackable, Personal and Human”: Google is modifying its approach to presenting search results, prioritizing conversations with AI, short videos and social media posts over traditional website results. This change is in response to evolving trends in information access, including the rise of AI like ChatGPT and the popularity of apps like TikTok. Google aims to create a personalized, visual and snackable experience that appeals to younger audiences, integrating human voices and user-generated content. As a result, Google’s users will see follow-up questions or browse through visuals such as TikTok videos. Ensuring the reliability of “trusted” content will be a challenge for Google. Website owners may be affected, but introducing conversational AI features should be well received. Although there is concern about fabricated information generated by chatbots like ChatGPT and the spread of user-generated content in search results, Google is refining its approach to addressing these issues. Dive deeper with this Search Engine Land article.
Google Bard Waitlist Removed, More Features Announced: Google has globally released the latest version of Bard, featuring expanded language support in US English, Japanese and Korean. The upgraded Google Bard presents a more richly visual interface with the support of Google Lens and integration with other features. Firstly, Google now uses data from Google Search and knowledge graphs to imbue the latest Bard with greater visual prowess and helpfulness. Google Maps and Google Lens have also been synchronized with Bard, providing users with more features and ease of access. Additional coding improvements have enabled source citations, a valuable “export” button and better coding assistance in Bard. Users can now enjoy “Dark mode,” making Bard even easier on the eyes, as well as the simple option to instantly export content into Google Docs and Gmail. Bard also integrates Adobe Firefly, enabling users to create images instantly and effortlessly. Check out Google’s Keyword article here to see more of these features.
Bing To Add “Learn More On” Section in Their Search Experience: Microsoft Bing is reportedly experimenting with a new “Learn more on” section that purportedly offers search refinements and expanders for related information. The search engine is also looking to augment its online retail game by featuring a box called “Microsoft Edge Deal Days” in a few search results that could entice users to shop online or via the Edge browser. This is anticipated to appear when ambiguous search terms require further elaboration. Additionally, Bing has introduced an icon and countdown overlay aimed at shopping ads on the search results page, which apparently indicates a focus on transactions and integration with ChatGPT, making Edge a pivotal product for Microsoft. Learn more about this news with this Search Engine Roundtable post.
Microsoft Ads for Chat API, Coming to AI Chat Platforms: Microsoft recently launched Ads for Chat API to monetize AI chat experiences for partners, advertisers and consumers. The API facilitates the creation of native advertising experiences, promoting AI-powered chatting and personalized product recommendations based on chats. These solutions can benefit select publishers, Microsoft Start partners and apps. The Ads for Chat API enables partners to customize advertisements, ensuring that ads align with the platform’s native environment, providing consumers with a less disruptive experience. Microsoft intends to cooperate with publishers, apps and online services that seek to monetize their AI chat functionality through advertising. The rollout of Ads for Chat API marks a significant development following recent Bing and Edge updates and the lifting of Bing’s AI waitlist. Microsoft is expected to reveal advancements in machine learning, GPT-4, Azure OpenAI Service, .NET and more during its upcoming Microsoft Build event on May 23.
Amazon Advertisers Will Soon Get Their Taste of AI-Generated Assets: Amazon is assembling a squad to deliver advertisers with AI-generated image and video resources. The timeline for their release remains uncertain, so we’ll wait and see how this would impact Amazon store owners. However, given their potential to improve listings and generate greater sales, there is undoubtedly cause for excitement. With Amazon’s total revenue generated from advertising projected to reach $38 billion by 2022, the company’s current primary offering, increased visibility for merchants in search results, remains a key focus. Read the full story in this post by Search Engine Land.
Editor’s Note: “SEO News You Can Use” is a weekly blog post posted every Monday morning only on SEOblog.com, rounding up all the top SEO news from around the world. Our goal is to make SEOblog.com a one-stop-shop for everyone looking for SEO news, education and for hiring an SEO expert with our comprehensive SEO agency directory.
