Despite recent narratives painting Google as trailing behind in the AI race, the tech giant's annual developer event served as a potent reminder of its deep-seated roots and continued leadership in the field. The company unveiled a multitude of AI enhancements designed to bolster its expansive portfolio of products and services. Detractors may argue that Google is lagging behind competitors such as Microsoft in certain domains, but these criticisms may overlook the broader narrative. Here are four critical takeaways from the event that paint a picture of Google's AI future:
PaLM 2 is not Google’s response to GPT-4, Gemini is
Looking back at Google's announcements this week, it is clear that PaLM 2 is not their response to GPT-4. Instead, it seems to be a cleverly orchestrated PR move to alleviate concerns among stakeholders regarding the growing momentum surrounding OpenAI and Microsoft. Google's announcement of domain-specific models like Med-PaLM 2 and Sec-PaLM isn't new; we've heard about these applications before. Furthermore, the purported improvements in multilingual, reasoning, and coding capabilities in PaLM 2 still fall significantly behind GPT-4's performance. In reality, the announcement served more as a reassurance of Google's ongoing efforts and less as a demonstration of competitive advantage.
Interestingly, the pre-announcement of Gemini, a completely new foundation model, rather than an advancement to PaLM 3, is highly suggestive. It reveals Google's current predicament – they're trying to match the strides OpenAI has made with GPT-4, but their existing framework isn't quite there yet. The introduction of Gemini signifies a new direction, a fresh start, a tacit admission that their current generation of AI can't compete head-to-head with GPT-4. Therefore, rather than viewing these announcements as a revelation of Google's advancements, it's more accurate to perceive them as a strategic move to maintain stakeholder confidence while they navigate the next phase of their AI journey.
Microsoft 365 Copilot has set a precedent that Google is now striving to match
Google’s announcement of Duet AI for Workspace and Google Cloud, is a tacit acknowledgement that Microsoft has one-upped them in responding to the rising demand for generative AI in the workplace. Over the last six months, co-creating with AI has rapidly transitioned from being a fascinating novelty to an essential requirement. In response, companies are accelerating their efforts to cater to this growing demand.
Duet AI mirrors many of the capabilities offered by Microsoft 365 Co-pilot. With Duet AI for Google Workspace, the search giant aims to facilitate a more seamless and efficient work environment, leveraging AI to streamline tasks such as scheduling, document management, and communication.
However, the more noteworthy announcement is Duet AI for Google Cloud. It is an unmistakable response to Microsoft's Github Copilot X. In spite of Microsoft's robust developer ecosystem, fueled by the acquisition of GitHub and the popularity of their developer tools like VS Code, Google is pushing back with strategic alliances. They announced new partnerships with Replit and Gitlab which signify a concerted effort to bolster their developer toolset and community. With Google's extensive resources backing these strategic alliances, they could certainly pose a formidable challenge to Microsoft’s head start.
Still, Microsoft's investment and partnership with OpenAI has given them a significant advantage. OpenAI’s GPT technology has not only enhanced the allure of Microsoft's suite of applications, but has helped them to position themselves as pioneers in incorporating AI with everyday work tools. It sets an intriguing stage for a technology rivalry, as Google strives to match Microsoft's pioneering stride in this space. Yet, the continual evolution of AI and its potential applications mean that the finish line is always shifting - this competition is far from over.
Google is reshaping search with generative AI, but Bard remains on the sidelines
Google's announcement of supercharging search with generative AI shows a cautious yet promising evolution in the tech giant's approach to search. Unlike Microsoft's Bing, which has quickly reimagined its search experience with the integration of Bing Chat as a core part of the user experience, Google seems to be taking a more measured approach. Their experimental AI, Bard, remains separate from the core search experience for now, despite the potential it offers. While Google is leveraging the power of generative AI to refine and enhance its search results, it is clear that they are mindful of maintaining a balance between introducing innovative technology and maintaining a familiar and trusted user experience.
The introduction of a new conversational mode, however, suggests that Google is not entirely averse to the idea of a more interactive search experience. This mode, which allows users to ask Google follow-up questions on the topic they're exploring, could be a precursor to a more integrated use of Bard in the future. It's also worth noting that Google's commitment to highlighting web content and maintaining dedicated ad slots signals a careful consideration of the ecosystem in which it operates. Google understands that ads are an integral part of the web infrastructure and play a crucial role in connecting people with relevant products and services. Thus, while the changes to Google's search experience might not be as radical as Microsoft's, they signify a thoughtful and balanced approach to innovation.
Google doesn’t need have the best foundation model to win
The number of times the word “AI” was mentioned on stage at Google I/O, underscores its pivotal role in Google's future business strategy. And while the unveiling of new foundational models and next-generation tools was indeed noteworthy, what stands out most is Google's strategic move to position itself as the core infrastructure and toolset provider for the AI-driven applications of tomorrow. Their intentions are clear: they aim to be a cornerstone of the AI-powered future rather than just a participant.
However, Google's approach extends beyond merely offering its own AI models. The company is strategically fostering an ecosystem of choice and ease-of-use, exemplified by Vertex AI and Duet AI for Google Cloud. In addition to offering its own models, Google is providing a range of popular open-source models and plans to accommodate third-party models in the near future. The end goal? To position Google Cloud as the go-to platform for building and hosting AI applications, as opposed to just the source of another AI model in what is likely to become a crowded market.
The brilliance of this strategy is already visible in the impressive list of leading AI companies that Google has successfully attracted as customers or partners, including names like Box, Canva, Dialpad, Jasper, Salesforce, AI21 Labs, Anthropic, Cohere, Midjourney, and Replit. This isn't just a smart play for today; it's a shrewd business move that anticipates the AI-driven landscape of tomorrow.