As expected, this year’s Google I/O developer conference was heavily focused on AI (Google A/I?). In a pre-briefing, Google told reporters that it aims to make “AI accessible for every developer.” Ergo, a raft of new or updated AI APIs and models were announced.
However, deep questions remain about how all this AI pixie dust will impact the wider web ecosystem — in particular, Google Search, a product that most web operators rely on to stay in business. After all, what good is AI in developer tools if those developers soon won’t have jobs?
Sprinkling AI on the Web Platform
Google has made a lot of AI additions to its web platform tooling.
From Chrome 126, Gemini Nano will be built into the Chrome desktop. Nano is Google’s “most efficient model for on-device tasks” among the Gemini models. According to Paul Kinlan, lead for Chrome and web developer relations at Google, embedding Nano into Chrome desktop means developers need not “worry about prompt engineering, fine-tuning, capacity, or cost.”
However, Nano itself will be able to be fine-tuned, said Josh Woodward, VP of Google Labs, at the press briefing.
One of Google’s strengths has traditionally been using advanced features of the web platform to engineer speed and performance improvements in its Chrome browser. With that in mind, Kinlan announced that Google has “invested heavily to ensure that AI models run quickly and efficiently by using WebGPU and Wasm — the backbone technologies that enable on-device AI on the web.” Together with Chrome, this kind of platform support for AI engineering is one area where Google can get a leg up over its AI rivals OpenAI, Microsoft and Meta.
Gemini featured in Project IDX.
In a less impactful update, but one sure to please developers, Google announced it is “bringing Gemini to the Chrome DevTools Console” to help devs with debugging. This was made available today as an experimental feature in the U.S., “rolling out to more countries soon.”
Various other AI tweaks have been made to Chrome, ostensibly to improve the browsing experience of users. Kinlan noted that “AI can be used to intelligently predict navigation patterns,” which combined with APIs such as the Speculation Rules API and the View Transitions API, can help with pre-fetching and pre-rendering of web pages in the background.
To help devs keep up with all these web developer changes, Google has released a new “Web Platform Dashboard,” which will enable developers “to see the entire web platform mapped out as a set of features, follow their development, and check their interop status.”
As well as the web platform additions, Google announced a bunch of other tweaks across the dev board — such as partial hydration in Angular, 3D immersive experiences in the Maps JavaScript API, and new versions of its mobile platforms Flutter and Dart.
Project IDX: Google’s Chosen AI IDE?
For those of us who have been waiting a while to try out Project IDX, Google’s cross-platform IDE, there was good news at I/O. Project IDX has now been released as a public beta, with no more waiting list.
In my interview with Project IDX lead Kirupa Chinnathambi last November, I explained that Project IDX is for developers who want to build an app that runs on both the web and on various mobile operating systems. It has similarities to GitHub Codespaces, in that both products integrate with Code OSS, Microsoft’s open source Visual Studio Code product. But Chinnathambi told me at the time that IDX is “more opinionated.” He noted that it is able to do Android emulation and iOS simulation, as one example.
And yes, Project IDX got more AI features today at I/O. “We’ve integrated the Gemini model deeply into IDX to provide you with assistance directly in your workspace,” Google stated.
AI assistant in Project IDX.
Project IDX already had support for AI-powered code completion, an assistive chat, and contextual code actions like “add comments” and “explain this code”. At I/O, we were simply told that these features have been “improved.”
At the press briefing, I asked if there has been any work done across Google’s developer products to help with web design processes — for example, creating CSS and layouts. These design-focused AI features are now common in tools like Figma and Locofy.
“We have plans this year to introduce more transformation capabilities and code fixing,” replied Jeanine Banks, VP & GM of Developers at Google. “And then I think the other important thing here is the connection to design as well. And so how can design workflows drive into code? And that’s a space that we’re looking at as well.”
Code completion.
Firebase Genkit
It wouldn’t be a developer conference without some kind of new framework being announced. Sure enough, Google is introducing Firebase Genkit — a new open source framework built for JavaScript/TypeScript developers (“with Go coming soon”) to help them create Node.js backends for AI applications.
You won’t have to use Firebase, however. Google notes that if you use VS Code or Project IDX, “you can also open the Genkit Developer UI in the VS Code integrated browser and use it side-by-side with your code.”
To start with, Genkit supports the following integrations:
- Large Language Models such as Google’s Gemini and open source models via Ollama.
- Vector Databases like Chroma, Pinecone, Cloud Firestore, and PostgreSQL (pgvector).
- Embedding Services from Google (Google AI & Vertex).
Search Impact on Web Ecosystem
Despite the raft of AI functionalities announced today at Google I/O, there was no mention of the impact of AI search on the web platform. While search is not usually the point of I/O for developers, there has been heightened interest this year in how Google will defend its search turf against an AI-enhanced Microsoft Bing, new entrants like Perplexity, and maybe even a “Search GPT” from OpenAI.
Many web developers work for companies whose websites or web apps would be under threat from an AI-powered Google search. So for Google I/O to not even bring up the existential threat of AI for the web… well, it was disappointing.
That being said, since the web ecosystem is not dead yet, I suppose it is useful for developers to have a bunch of shiny new AI tools.
The post Devs Get AI Pixie Dust at Google I/O — But AI Search Impact? appeared first on The New Stack.
This year’s Google I/O developer conference was heavily focused on AI. However, web ecosystem concerns about AI and Google Search went unanswered.