During Google’s I/O developer conference, the company highlighted its advancements in AI to both developers and, to a certain extent, consumers. Presenting various innovations, Google introduced a new AI-powered search engine, announced an AI model featuring an expanded context window of 2 million tokens, and showcased AI assistants integrated into its Workspace suite, including Gmail, Drive, and Docs. Additionally, Google revealed tools to enable developers to embed its AI into their applications and offered a glimpse into its future AI vision, known as Project Astra, which can interpret and respond to visual, auditory, vocal, and textual inputs.
While each individual announcement was promising, the sheer volume of AI news was overwhelming. Though targeted primarily at developers, these high-profile events also serve to captivate the interest of end-users. However, the flood of information may leave even tech-savvy consumers puzzled, questioning: what exactly is Project Astra? Is it the technology behind Gemini Live? Is Gemini Live similar to Google Lens? How does it differ from Gemini Flash? Is Google truly developing AI glasses or are those just speculative? What’s Gemma, and where does LearnLM fit in? What are Gems? When will Gemini be available in inboxes or document applications? How are these technologies to be utilized effectively?
Despite the palpable enthusiasm from individual presenters and the enthusiastic cheers from Google employees, the presentation lacked a clear indication of an imminent AI revolution. If artificial intelligence is set to profoundly influence the trajectory of technology, akin to the iPhone’s impact on personal computing, this event did not mark its grand unveiling.
The overarching message was that we are still in the nascent stages of AI development.
On the event’s periphery, there was an evident acknowledgment among Googlers that their work remained unfinished. For instance, during a demonstration showcasing AI’s ability to generate a student’s study guide and quiz in mere moments from a multi-hundred-page document – a notably impressive feat – it was evident that the answers were not cross-referenced with cited sources. When questioned about accuracy, an employee conceded that the AI generally gets answers right and that future iterations would include source citations for fact-checking. However, if fact-checking is necessary, it raises questions about the reliability of an AI-generated study guide in test preparation.
In another demonstration called Astra, a camera mounted over a table connected to a large touchscreen enabled users to engage in activities like playing Pictionary with the AI, inquiring about objects, getting storytelling assistance, and more. Despite the evident technical sophistication, the practical applications of these capabilities in everyday life were not immediately clear.
For example, the AI could use alliteration to describe objects. During the livestreamed keynote, Astra identified crayons and responded with “creative crayons colored cheerfully.” An amusing party trick, perhaps.
In a private demo, when the AI was prompted to identify objects in hastily drawn sketches, it successfully recognized a flower and a house on the touchscreen. However, when presented with a simple sketch of a bug—comprising a large circle for the body, a smaller circle for the head, and little legs—the AI struggled. It guessed incorrectly several times until the employee provided guidance, leading the AI to identify it as something living. After adding two more legs to make it an eight-legged figure, the AI finally identified it as a spider. A human observer would have immediately recognized the basic drawing despite its simplicity.
To provide a clearer understanding of the current state of technology, it should be noted that during the Astra demo, Google staff prohibited both recording and photography. Additionally, while Astra was operational on an Android smartphone, attendees were neither allowed to view the app nor handle the device. Although the demonstrations were engaging and showcased intriguing technological advancements, Google missed a crucial opportunity to illustrate how its AI technology could be integrated into everyday life.
Consider the practicality of some of the showcased applications: Is it truly necessary to enlist an AI to generate a band name from an image of your dog and a stuffed tiger? How essential is AI assistance in locating your glasses? These, among other demos featured during the keynote, may not convincingly demonstrate how the technology will be relevant to everyday users.
In recent times, we’ve repeatedly seen technology events showcasing futuristic advancements that often lack immediate real-world applications or present minor conveniences as major innovations. A prime example is Google, which has previously showcased its AR glasses. Notably, there was a dramatic demonstration where skydivers descended into I/O wearing Google Glass—a project initiated over a decade ago but ultimately discontinued.
Upon reviewing this year’s I/O, it becomes apparent that Google views AI primarily as a revenue-generating tool, evidenced by offerings like Google One AI Premium for enhanced products. This indicates a possible reluctance from Google to spearhead a revolutionary consumer AI breakthrough. Echoing this sentiment, OpenAI’s CEO Sam Altman recently reflected on their mission, stating that the initial goal was to advance technology for widespread global benefit. He noted, “Instead, it now looks like we’ll create AI and then other people will use it to create all sorts of amazing things that we all benefit from.”
Google seems to be on a similar trajectory.
However, there were moments when Google’s Astra AI appeared particularly promising. Its capacity to accurately identify code or provide system enhancement suggestions based on diagrams underscores its potential as a valuable work companion—think of it as an advanced iteration of Clippy.
There were several instances where the tangible benefits of AI in real-world applications were evident. Integrating Gemini’s AI into your email could assist in summarizing messages, drafting replies, and listing tasks, which might help achieve the elusive “inbox zero” or get closer to it more expediently. However, its current capabilities fall short of efficiently sorting non-spam but unwanted emails, intelligently organizing emails into categories, ensuring you never overlook an important message, and providing an immediate overview of actionable items upon login. Additionally, summarizing critical news from email newsletters is still beyond its reach.
When contemplating the impact of AI on the Android ecosystem—Google’s core message to attending developers—there was an underlying realization that even Google has yet to convincingly argue that AI will shift users from Apple’s ecosystem to Android. When asked, “When is the best time to switch from iPhone to Android?”, various Google representatives consistently responded with, “This fall.” This timeframe aligns with Google’s autumn hardware event, which is expected to coincide with Apple’s support for RCS—an enhancement to SMS that aims to make Android’s messaging experience more competitive with iMessage.
In essence, the widespread adoption of AI in personal technology may necessitate advancements in hardware—perhaps AR glasses, an enhanced smartwatch, or Gemini-powered Pixel Buds. Nevertheless, Google remains reticent about disclosing its forthcoming hardware innovations or even hinting at them. The lukewarm receptions of products like the AI Pin and Rabbit underscore that hardware development continues to be a challenging endeavor.
Despite significant advancements in Google’s AI capabilities for Android devices, its recent I/O event left much to be desired concerning other key products. The Pixel Watch and its operating system, WearOS, received only minor performance enhancements and scant attention. Additionally, Google’s Pixel Buds were conspicuously absent from the conversation. In contrast, Apple’s ecosystem tightly integrates accessories like the Apple Watch and AirPods, which are pivotal to its user retention strategy and increasingly interwoven with AI, potentially powered by Siri.
Anticipation now shifts towards Apple’s Worldwide Developer Conference (WWDC), where the tech giant is expected to reveal its AI strategy. Speculation abounds about potential collaborations, possibly involving OpenAI or, intriguingly, even Google. The critical question is whether Apple’s AI will achieve seamless OS integration comparable to Google’s Gemini on Android.
As Google’s fall hardware event approaches, it has a strategic window to assess Apple’s innovations and respond with a compelling AI narrative. The company seeks a breakthrough moment reminiscent of Steve Jobs’ iconic iPhone introduction, encapsulating a clear and immediate understanding of its AI prowess.
While this year’s I/O fell short of providing that clarity, the industry and consumers alike remain eager to see when and how Google will deliver its AI vision with the same impact and coherence.
