Image Credits: David Paul Morris/Bloomberg / Getty Images
Google’s upgrading its image-generation tech to keep apace with rivals.
At the company’s I/O developer conference in Mountain View on Tuesday, Google announced Imagen 3, the latest in the tech giant’s Imagen generative AI model family.
Demis Hassabis, CEO of DeepMind, Google’s AI research division, said that Imagen 3 more accurately understands the text prompts that it translates into images versus its predecessor, Imagen 2, and is more “creative and detailed” in its generations. In addition, the model produces fewer “distracting artifacts” and errors, he said.
“This is [also] our best model yet for rendering text, which has been a challenge for image-generation models,” Hassabis added.
To allay concerns around the potential to create deepfakes, Google says that Imagen 3 will use SynthID, an approach developed by DeepMind to apply invisible, cryptographic watermarks to media.
Sign-ups for Imagen 3 in private preview are available in Google’s ImageFX tool, and Google says the model will “come soon” to devs and corporate customers using Vertex AI, Google’s enterprise generative AI development platform.
Image Credits: Google
Google typically doesn’t reveal much about the source of the data it uses to train its AI models — and this time was no exception. There’s a reason for that. Much of the training data comes from public sites, repositories and datasets around the web. And some of that training data, specifically the copyrighted data scraped without permission from content creators, is a source of IP-related lawsuits.
Google’s web publisher controls allow webmasters to prevent the company from scraping data, including photos and videos, from their websites. But Google doesn’t offer an “opt-out” tool, and — unlike some of its rivals — the company hasn’t committed to compensating rights holders for their (in some cases unknowing) contributions to the training datasets.
The lack of transparency isn’t surprising. But it is disappointing — especially from a company with resources like Google’s.
We’re launching an AI newsletter! Sign up here to start receiving it in your inboxes on June 5.
Image Credits: Alex Tai/SOPA Images/LightRocket / Getty Images
Gemini model capabilities are coming to the Google Maps Platform for developers, starting with the Places API, the company announced at the Google I/O 2024 conference on Tuesday. With this new capability, developers will be able to show generative AI summaries of places and areas in their own apps and websites.
The summaries are created based on Gemini’s analysis of insights from Google Maps’ community of more than 300 million contributors. With this new capability, developers will no longer have to write their own custom descriptions of places.
For example, if a developer has a restaurant-booking app, this new capability will help users understand which restaurant is best for them. When users search for restaurants in the app, they’ll be able to quickly see all the most important information, like the house specialty, happy hour deals and the place’s vibes.
Image Credits: Google
The new summaries are available for many types of places, including restaurants, shops, supermarkets, parks and movie theaters.
Google is also bringing AI-powered contextual search results to the Places API. When users search for places in a developer’s product, developers will now be able to display reviews and photos related to their search.
If a developer has an app that allows users to explore local restaurants, their users can search “dog-friendly restaurants,” for example, to see a list of relevant dining spots, along with relevant reviews and photos of dogs at the restaurants.
Contextual search results are available globally, and place and area summaries are available in the U.S. Google plans to expand them to more countries in the future.
We’re launching an AI newsletter! Sign up here to start receiving it in your inboxes on June 5.
Developer season has officially commenced, and we’re just one day away today, Google I/O 2024 kicks off in Mountain View’s Shoreline Amphitheatre. The short answer to the question of what to expect is two letters: AI. Google I/O? More like Google A/I. There’s been an element of artificial intelligence/machine learning in almost every Google announcement for the past several years, but this time, you’re going to be sick of the subject by the time CEO Sundar Pichai leaves the stage.
Things kick off today, May 14 at 10 a.m. PT. That’s the big keynote — the one you think of when you think about the event. “Discover how we’re furthering our mission to organize the world’s information and make it universally accessible and useful,” Google says. It’s the keynote that famously once involved skydivers shilling a breakthrough piece of augmented reality called Google Glass that we all wear every day.
A few hours later, at 1:30 p.m. PT is the Developer keynote. Google describes that one like this: “Learn about Google’s newest developer tools and discover how they fuel innovation and enhance your productivity.” Might be a bit wonky, especially if you have no development skills to speak of.
For the sake of this article, we’re going to be focused on the former.
Google Pixel Family BundleImage Credits: Google
New Pixel devices likely won’t be present at the event, however. Unlike past years when the company gave over some stage time to a new handset, Google spoiled the surprise, revealing the Pixel 8a earlier this week. It was a curious decision, especially given that the budget device officially starts shipping the day of the keynote.
While the Pixel Fold was a star of last year’s I/O, the current round of rumors suggest that the foldable is about to undergo a major rebrand. Rather than Pixel Fold 2, we will reportedly be getting the far more unwieldy Pixel 9 Pro Fold. It’s the kind of overlong name you need a fold-out screen to appreciate. That branding would likely find the device being announced in the fall, along with the other Pixel 9s.
Image Credits: Google
Will there be any hardware at all? Perhaps we’ll see an update on some previous concepts. For instance, what ever happened to those translation glasses? Or how about Project Starline? I got a demo of the slick experimental teleconferencing system at last year’s event, but things seem to have gone silent in the intervening 12 months.
The Nest line has been neglected for a while now. Perhaps Google has backed away from the smart home business as it has begun to cool. Or maybe the company is ramping up for something new. Something we would love to hear about at the presser is what Google Gemini’s arrival means for Google Assistant. Is the former outright replacing the latter? Can they peacefully coexist?
Image Credits: Lorenzo Di Cola/NurPhoto / Getty Images
Whether Google broaches the Assistant conversation, I can say with certainty that Gemini will be center stage. The company has a lot riding on its generative AI platform. It will need to demonstrate that it has surpassed chief competitors like OpenAI’s GPT and Microsoft Copilot. It’s plausible that we could see a wholly new version of Gemini next week. We’ll also see the platform more deeply integrated into services like Maps and Chrome, as well as Android.
A developer preview of Android 15 reared its head last month. Google could well offer a deeper dive into its next mobile operating system. Expect to see some key new features for the OS, which is set to feature deep Gemini integration.
We’ll no doubt be getting previews of new versions of Google’s other platforms, including Wear OS, Android TV and Android Auto. More AI developer tools will almost certainly be on the docket for the event, as well.
Maybe we’ll even get another cameo appearance from our friends in the protest plane. Things kick off at 10 a.m. PT today. You can stream the event here and stay tuned to TechCrunch for news as it drops.
Back in February, Google paused its AI-powered chatbot Gemini’s ability to generate images of people after users complained of historical inaccuracies. Told to depict “a Roman legion,” for example, Gemini would show an anachronistic group of racially diverse soldiers while rendering “Zulu warriors” as stereotypically Black.
Google CEO Sundar Pichai apologized, and Demis Hassabis, the co-founder of Google’s AI research division DeepMind, said that a fix should arrive “in very short order” — within the next couple of weeks. But we’re now well into May, and the promised fix remains elusive.
Google touted plenty of other Gemini features at its annual I/O developer conference this week, from custom chatbots to a vacation itinerary planner and integrations with Google Calendar, Keep and YouTube Music. But image generation of people continues to be switched off in Gemini apps on the web and mobile, confirmed a Google spokesperson.
So what’s the holdup? Well, the problem’s likely more complex than Hassabis alluded to.
The datasets used to train image generators like Gemini’s generally contain more images of white people than people of other races and ethnicities, and the images of non-white people in those datasets reinforce negative stereotypes. Google, in an apparent effort to correct for these biases, implemented clumsy hardcoding under the hood. And now it’s struggling to suss out some reasonable middle path that avoids repeating history.
Will Google get there? Perhaps. Perhaps not. In any event, the drawn-out affair serves as a reminder that no fix for misbehaving AI is easy — especially when bias is at the root of the misbehavior.
We’re launching an AI newsletter! Sign up here to start receiving it in your inboxes on June 5.
Netflix announced during its Upfronts presentation on Wednesday that it’s launching its own advertising technology platform only a year and a half after entering the ads business. This move pits it against other industry heavyweights with ad servers, like Google, Amazon and Comcast.
The announcement signifies a significant shake-up in the streaming giant’s advertising approach. The company originally partnered with Microsoft to develop its ad tech, letting Netflix enter the ad space quickly and catch up with rivals like Hulu, which has had its own ad server for over a decade.
With the launch of its in-house ad tech, Netflix is poised to take full control of its advertising future. This strategic move will empower the company to create targeted and personalized ad experiences that resonate with its massive user base of 270 million subscribers.
“Bringing our ad tech in-house will allow us to power the ads plan with the same level of excellence that’s made Netflix the leader in streaming technology today,” said Amy Reinhard, Netflix’s president of advertising. “We’re being incredibly strategic about how we present ads because we want our members to have a phenomenal experience. We conduct deep consumer research to make sure we stay ahead of the competition, bringing opportunities that are better for members and better for brands.”
Netflix didn’t say exactly how its in-house solution will change the way ads are delivered, but it’s likely it’ll move away from generic advertisements. According to the Financial Times, Netflix wants to experiment with “episodic” campaigns, which involve a series of ads that tell a story rather than delivering repetitive ads.
During the presentation, Netflix also noted that it’ll expand its buying capabilities this summer, which will now include The Trade Desk, Google’s Display & Video 360 and Magnite as partners. Notably, competitor Disney+ also has an advertising agreement with The Trade Desk.
Netflix also touted the success of its ad-supported tier, reporting that 40 million global monthly active users opt for the plan. The ad tier had around 5 million users within six months of launching.
Despite challenges, Netflix says its ad tier is doing well
Image Credits: Google / Google CEO Sundar Pichai at the Google I/O 2024 event
It’s that moment you’ve been waiting for all year: Google I/O keynote day! Google kicked off its developer conference each year with a rapid-fire stream of announcements, including many unveilings of recent things it’s been working on. Brian already kicked us off by sharing what we are expecting.
Since you might not have had time to watch the whole two-hour presentation Tuesday, we took that on and delivered quick hits of the biggest news from the keynote as they were announced, all in an easy-to-digest, easy-to-skim list.
Google I/O takeaways
It was an AI evolution, not revolution
The company made its case to developers — and to some extent, consumers — why its bets on AI are ahead of rivals. At the event, the company unveiled a revampedAI-powered search engine, an AI model with an expanded context window of 2 million tokens, AI helpers across its suite of Workspace apps, like Gmail, Drive and Docs, tools to integrate its AI into developers’ apps and even a future vision for AI, codenamed Project Astra, which can respond to sight, sounds, voice and text combined.
While each advance on its own was promising, the onslaught of AI news was overwhelming. Though obviously aimed at developers, these big events are also an opportunity to wow end users about the technology. But after the flood of news, even somewhat tech-savvy consumers may be asking themselves, wait, what’s Astra again? Is it the thing powering Gemini Live? Is Gemini Live sort of like Google Lens? How is it different from Gemini Flash? Is Google actually making AI glasses or is that vaporware? What’s Gemma, what’s LearnLM…what are Gems? When is Gemini coming to your inbox, your docs? Read more
The top new AI products and features unveiled
From generative AI to accessibility, Kyle Wiggers takes you on a journey of all of Google’s AI announcements. Read more
Privacy concerns over AI voice call scans
Google showcased a demo of a call scam detection feature during I/O, which it says will be added to a future version of Android. The feature scans voice calls as they’re happening with AI, which effectively is client-side scanning, which has already sparked such a backlash on iOS that Apple abandoned its plans to adopt it in 2021. And as expected, a number of privacy advocates and experts voiced concerns over Google’s use of the technology, which they say could swiftly expand beyond applying just to scams and be used in more malicious ways. Read more
Updated security features
On Wednesday, Google announced it is adding new security and privacy protections to Android, including on-device live threat detection to catch malicious apps, new safeguards for screen sharing, and better security against cell site simulators.
The company said it is increasing the on-device capability of its Google Play Protect system to detect fraudulent apps trying to breach sensitive permissions. It also uses AI to detect if apps are trying to interact with other services and apps in an unauthorized manner.
Google said if the system is certain about malicious behavior, it disables the app automatically. Otherwise, it alerts the company for a review and then alerts users. Read more
And to protect devices in the real world, Google also announced Theft Detection Lock, an AI-powered addition that identifies motion commonly associated with theft, like a swift movement in an opposite direction. Once detected, the phone screen automatically locks, preventing future usage of the device without clearing whatever safeguards you’ve put in place. Read more
Google TV
Image Credits: Google
Google worked its Gemini into its Google TV smart TV operating system so it can generate descriptions for movies and TV shows. When a description is missing on the home screen, the AI will fill it in automatically to ensure that viewers never have to wonder what a title is about. It’ll also translate descriptions into the viewer’s native language, making the content more discoverable to a wider audience. The best part? The AI-generated descriptions are also personalized based on a viewer’s genre and actor preferences. Read more
Private Space feature
Image Credits: Google
Now here’s a fun one. Private Space is a new Android feature that lets users silo a portion of the operating system for sensitive information. It’s a bit like Incognito mode for the mobile operating system, sectioning designated apps into a “container.”
The space is available from the launcher and can be locked as a second layer of authentication. Apps in Private Space will be hidden from notifications, settings and recents. Users can still access the apps through a system sharesheet and photo picker in the main space, so long as the private space has been unlocked.
Developers can play around with it now, but there is a caveat — there is a bug. Google says it expects to address the bug in the coming days. Read more
Google Maps gets geospatial AR
Google Maps users will soon have a new layer of content on their phones — they will have access to geospatial augmented reality content. The feature will first appear in Singapore and Paris as part of a pilot program launching later this year.
Users will be able to access the AR content by first searching for a location in Google Maps. If the location has AR content and the user is near the place, they will have to tap on the image that says “AR Experience” and then lift their phone.
If someone is exploring a place remotely, they can see the same AR experience in Street View. After exploring the AR content, users can share the experience through a deep link URL or QR code on social media. Read more
Wear OS 5
Image Credits: Google
Google gave a developer preview of the new version of its smartwatch operating system, Wear OS 5. The latest release focuses on improved battery life and other performance improvements, like more efficient workout tracking. Developers are also getting updated tools for creating watch faces, as well as new versions of Wear OS tiles and Jetpack Compose for building watch apps. Read more
TechCrunch Minute
As we note all over this post, the Google I/O developer conference came with a big dose of AI. See how Anthony Ha summed it up Wednesday. Read more
Even Elon Musk took note
👌 https://t.co/I83VTEPBMy
— Elon Musk (@elonmusk) May 15, 2024
“Web” search filter
We’ve launched a new “Web” filter that shows only text-based links, just like you might filter to show other types of results, such as images or videos. The filter appears on the top of the results page alongside other filters or as part of the “More” option, rolling out today… pic.twitter.com/tIUy9LNCy5
— Google SearchLiaison (@searchliaison) May 14, 2024
Google introduced a new way to filter for just text-based links. The new “Web” filter appears at the top of the results page and enables users to filter for text links the way they can today filter for images, video, news or shopping.
As Sarah Perez reports, the launch is an admission that sometimes people will want to just surface text-based links to web pages, aka the classic blue links, that today are often of secondary importance as Google either answers the question in its informational Knowledge Panels or, now, through AI experiments. Read more
Firebase Genkit
Image Credits: TechCrunch
There’s a new addition to the Firebase platform, called Firebase Genkit, that aims to make it easier for developers to build AI-powered applications in JavaScript/TypeScript, with Go support coming soon. It’s an open source framework, using the Apache 2.0 license, that enables developers to quickly build AI into new and existing applications.
Some of the use cases for Genkit the company is highlighting Tuesday include many of the standard GenAI use cases: content generation and summarization, text translation and generating images. Read more
AI ad nauseam
Of course Google used Gemini to count AI mentions during today's AI-filled #GoogleIO. And there was even one more after this. pic.twitter.com/ajL7JYPQVE
— TechCrunch (@TechCrunch) May 14, 2024
Tuesday’s Google I/O ran for 110 minutes, but Google managed to reference AI a whopping 121 times during (by its own count) the event. CEO Sundar Pichai referenced the figure to wrap up the presentation, cheekily stating that the company was doing the “hard work” of counting for us. Again, it was no surprise — we were ready for it. Read more
Generative AI for learning
Image Credits: Google
Also today, Google unveiled LearnLM, a new family of generative AI models “fine-tuned” for learning. It’s a collaboration between Google’s DeepMind AI research division and Google Research. LearnLM models are designed to “conversationally” tutor students on a range of subjects, Google says.
Though it is already available on several of Google’s platforms, the company is taking LearnLM through a pilot program in Google Classroom. It is also working with educators to see how LearnLM might simplify and improve the process of lesson planning. LearnLM could help teachers discover new ideas, content and activities, Google says, or find materials tailored to the needs of specific student cohorts. Read more
Quiz master
Image Credits: Google
Speaking of education, new to YouTube are AI-generated quizzes. This new conversational AI tool allows users to figuratively “raise their hand” when watching educational videos. Viewers can ask clarifying questions, get helpful explanations or take a quiz on the subject matter.
This is going to be some relief for those who have to watch longer educational videos, such as lectures or seminars, due to the Gemini model’s long-context capabilities. These new features are rolling out to select Android users in the U.S. Read more
Gemma 2 updates
Image Credits: Google
One of the top requests Google heard from developers is for a bigger Gemma model, so Google will be adding a new 27-billion-parameter model to Gemma 2. This next generation of Google’s Gemma models will launch in June. This size is optimized by Nvidia to run on next-generation GPU and can run efficiently on a single TPU host and vertex AI, Google said. Read more
Google Play is getting some attention with a new discovery feature for apps, new ways to acquire users, updates to Play Points and other enhancements to developer-facing tools like the Google Play SDK Console and Play Integrity API, among other things.
Of particular interest to developers is something called the Engage SDK, which will introduce a way for app makers to showcase their content to users in a full-screen, immersive experience that’s personalized to the individual user. Google says this isn’t a surface that users can see at this time, however. Read more
Detecting scams during calls
Image Credits: Google
Tuesday, Google previewed a feature it believes will alert users to potential scams during the call.
The feature, which will be built into a future version of Android, utilizes Gemini Nano, the smallest version of Google’s generative AI offering, which can be run entirely on-device. The system effectively listens for “conversation patterns commonly associated with scams” in real time.
Google gives the example of someone pretending to be a “bank representative.” Common scammer tactics like password requests and gift cards will also trigger the system. These are all pretty well understood to be ways of extracting your money from you, but plenty of people in the world are still vulnerable to these sorts of scams. Once set off, it will pop up a notification that the user may be falling prey to unsavory characters. Read more
Ask Photos
Image Credits: TechCrunch
Google Photos is getting an AI infusion with the launch of an experimental feature, Ask Photos, powered by Google’s Gemini AI model. The new addition, which rolls out later this summer, will allow users to search across their Google Photos collection using natural language queries that leverage an AI’s understanding of their photo’s content and other metadata.
While before users could search for specific people, places, or things in their photos, thanks to natural language processing, the AI upgrade will make finding the right content more intuitive and less of a manual search process.
And the example was cute, too. Who doesn’t love a tiger stuffed animal/Golden Retriever band duo called “Golden Stripes?” Read more
All About Gemini
Image Credits: Sarah Perez
Gemini in Gmail
Gmail users will be able to search, summarize, and draft their emails using its Gemini AI technology. It will also be able to take action on emails for more complex tasks, like helping you process an e-commerce return by searching your inbox, finding the receipt and filling out an online form. Read more
Image Credits: TechCrunch
Gemini 1.5 Pro
Another upgrade to the generative AI is that Gemini can now analyze longer documents, codebases, videos and audio recordings than before.
In a private preview of a new version of Gemini 1.5 Pro, the company’s current flagship model, it was revealed that it can take in up to 2 million tokens. That’s double the previous maximum amount. With that level, the new version of Gemini 1.5 Pro supports the largest input of any commercially available model. Read more
Gemini Live
The company previewed a new experience in Gemini called Gemini Live, which lets users have “in-depth” voice chats with Gemini on their smartphones. Users can interrupt Gemini while the chatbot’s speaking to ask clarifying questions, and it’ll adapt to their speech patterns in real time. And Gemini can see and respond to users’ surroundings, either via photos or video captured by their smartphones’ cameras.
At first glance, Live doesn’t seem like a drastic upgrade over existing tech. But Google claims it taps newer techniques from the generative AI field to deliver superior, less error-prone image analysis — and combines these techniques with an enhanced speech engine for more consistent, emotionally expressive and realistic multi-turn dialogue. Read more
Gemini Nano
Now for a tiny announcement. Google is also building Gemini Nano, the smallest of its AI models, directly into the Chrome desktop client, starting with Chrome 126. This, the company says, will enable developers to use the on-device model to power their own AI features. Google plans to use this new capability to power features like the existing “help me write” tool from Workspace Lab in Gmail, for example. Read more
Image Credits: Google
Gemini on Android
Google’s Gemini on Android, its AI replacement for Google Assistant, will soon be taking advantage of its ability to deeply integrate with Android’s mobile operating system and Google’s apps. Users will be able to drag and drop AI-generated images directly into their Gmail, Google Messages and other apps. Meanwhile, YouTube users will be able to tap “Ask this video” to find specific information from within that YouTube video, Google says. Read more
Image Credits: Google
Gemini on Google Maps
Gemini model capabilities are coming to the Google Maps platform for developers, starting with the Places API. Developers can show generative AI summaries of places and areas in their own apps and websites. The summaries are created based on Gemini’s analysis of insights from Google Maps’ community of more than 300 million contributors. What’s better? Developers will no longer have to write their own custom descriptions of places. Read more
Tensor Processing Units get a performance boost
Google unveiled its next generation — the sixth, to be exact — of its Tensor Processing Units (TPU) AI chips. Dubbed Trillium, they will launch later this year. If you recall, announcing the next generation of TPUs is something of a tradition at I/O, even as the chips only roll out later in the year.
These new TPUs will feature a 4.7x performance boost in compute performance per chip when compared to the fifth generation. What’s maybe even more important, though, is that Trillium features the third generation of SparseCore, which Google describes as “a specialized accelerator for processing ultra-large embeddings common in advanced ranking and recommendation workloads.” Read more
AI in search
Google is adding more AI to its search, assuaging doubts that the company is losing market share to competitors like ChatGPT and Perplexity. It is rolling out AI-powered overviews to users in the U.S. Additionally, the company is also looking to use Gemini as an agent for things like trip planning. Read more
Google plans to use generative AI to organize the entire search results page for some search results. That’s in addition to the existing AI Overview feature, which creates a short snippet with aggregate information about a topic you were searching for. The AI Overview feature becomes generally available Tuesday, after a stint in Google’s AI Labs program. Read more
Generative AI upgrades
Image Credits: Google
Google announced Imagen 3, the latest in the tech giant’s Imagen generative AI model family.
Demis Hassabis, CEO of DeepMind, Google’s AI research division, said that Imagen 3 more accurately understands the text prompts that it translates into images versus its predecessor, Imagen 2, and is more “creative and detailed” in its generations. In addition, the model produces fewer “distracting artifacts” and errors, he said.
“This is [also] our best model yet for rendering text, which has been a challenge for image generation models,” Hassabis added. Read more
Project IDX
Project IDX, the company’s next-gen, AI-centric browser-based development environment, is now in open beta. With this update comes an integration with the Google Maps Platform into the IDE, helping add geolocation features to its apps, as well as integrations with the Chrome Dev Tools and Lighthouse to help debug applications. Soon, Google will also enable deploying apps to Cloud Run, Google Cloud’s serverless platform for running front- and back-end services. Read more
Veo
Google’s gunning for OpenAI’s Sora with Veo, an AI model that can create 1080p video clips around a minute long given a text prompt. Veo can capture different visual and cinematic styles, including shots of landscapes and time lapses, and make edits and adjustments to already-generated footage.
It also builds on Google’s preliminary commercial work in video generation, previewed in April, which tapped the company’s Imagen 2 family of image-generating models to create looping video clips. Read more
Circle to Search
Image Credits: Google
The AI-powered Circle to Search feature, which allows Android users to get instant answers using gestures like circling, will now be able to solve more complex problems across psychics and math word problems. It’s designed to make it more natural to engage with Google Search from anywhere on the phone by taking some action — like circling, highlighting, scribbling or tapping. Oh, and it’s also better to help kids with their homework directly from supported Android phones and tablets. Read more
Pixel 8a
Image Credits: Google
Google couldn’t wait until I/O to show off the latest addition to the Pixel line and announced the new Pixel 8a last week. The handset starts at $499 and ships Tuesday. The updates, too, are what we’ve come to expect from these refreshes. At the top of the list is the addition of the Tensor G3 chip. Read more
Pixel Slate
Image Credits: Brian Heater
Google’s Pixel Tablet, called Slate, is now available. If you recall, Brian reviewed the Pixel Tablet around this time last year, and all he talked about was the base. Interestingly enough, the tablet is available without it. Read more
We’ll be updating this post throughout the day …
We’re launching an AI newsletter! Sign up here to start receiving it in your inboxes on June 5.
Google introduced a ton of new Gemini-related products at its Google I/O developer conference recently, and its Education suite is also getting an AI boost.
The company is launching a Gemini add-on for educational institutes through Google Workspace. There’s a base plan called Gemini Education that’s available in English for educators and students over 18, which lets users tap Gemini in Docs, Sheets, Slides and Meet, and chat with the AI.
And then you have Gemini Education Premium, which, in addition to everything in the base plan, provides features like AI-powered note-taking and summaries in Meet, as well as data loss prevention. The company said the base tier is priced at $24 per month or $192 per year per user. The premium plan costs $36 per month or $288 per year per user.
Image Credits: Google
Students above 18 at educational institutes can use Gemini for free without either of these plans.
Image Credit: Google
The company said Gemini comes with new education-focused features. The AI can explain topics step-by-step, help students practice the material they’re learning and can tap data from sources like Rice University’s OpenStax textbooks. Educators and instructors could use Gemini to create lesson plan templates, differentiate content to suit students, summarize research, personalize feedback and more.
During Google I/O, the company also released LearnLM, a new set of generative AI models aimed at the education sector. The company said LearnLM is already present in products like YouTube, Google’s Gemini apps, Google Search and Google Classroom. Additionally, the company said that it is adding AI-powered quizzing to academic videos on YouTube.
We’re launching an AI newsletter! Sign up here to start receiving it in your inboxes on June 5.
At Google’s I/O developer conference, the company made its case to developers — and to some extent, consumers — why its bets on AI are ahead of rivals. At the event, the company unveiled a revampedAI-powered search engine, an AI model with an expanded context window of 2 million tokens, AI helpers across its suite of Workspace apps, like Gmail, Drive and Docs, tools to integrate its AI into developers’ apps and even a future vision for AI, codenamed Project Astra, which can respond to sight, sounds, voice and text combined.
While each advance on its own was promising, the onslaught of AI news was overwhelming. Though obviously aimed at developers, these big events are also an opportunity to wow end users about the technology. But after the flood of news, even somewhat tech-savvy consumers may be asking themselves, wait, what’s Astra again? Is it the thing powering Gemini Live? Is Gemini Live sort of like Google Lens? How is it different from Gemini Flash? Is Google actually making AI glasses or is that vaporware? What’s Gemma, what’s LearnLM…what are Gems? When is Gemini coming to your inbox, your docs? How do I use these things?
If you know the answers to those, congratulations, you’re a TechCrunch reader. (If you don’t, click the links to get caught up.)
Image Credits: Google
What was missing from the overall presentation, despite the enthusiasm from the individual presenters or the whooping cheers from the Google employees in the crowd, was a sense of the coming AI revolution. If AI will ultimately lead to a product that will profoundly impact the direction of technology the way the iPhone impacted personal computing, this was not the event where it debuted.
Instead, the takeaway was that we’re still very much in the early days of AI development.
On the sidelines of the event, there was a sense that even Googlers knew the work was unfinished. When demoing how AI could compile a student’s study guide and quiz within moments of uploading a multihundred-page document — an impressive feat — we noticed that the quiz answers weren’t annotated with the sources cited. When asked about accuracy, an employee admitted that the AI gets things mostly right and a future version would point to sources so people could fact-check its answers. But if you have to fact-check, then how reliable is an AI study guide in preparing you for the test in the first place?
In the Astra demo, a camera mounted over a table and linked to a large touchscreen let you do things like play Pictionary with the AI, show it objects, ask questions about those objects, have it tell a story and more. But the use cases for how these abilities will apply to everyday life weren’t readily apparent, despite the technical advances that, on their own, are impressive.
For example, you could ask the AI to describe objects using alliteration. In the livestreamed keynote, Astra saw a set of crayons and responded “creative crayons colored cheerfully.” Neat party trick.
When we challenged Astra in a private demo to guess the object in a scribbled drawing, it correctly identified the flower and house I drew on the touchscreen right away. When I drew a bug — one bigger circle for the body, one smaller circle for the head, little legs off the sides of the big circle — the AI stumbled. Is it a flower? No. Is it the sun? No. The employee guided the AI to guess something that was alive. I added two more legs for a total of eight. Is it a spider? Yes. A human would have seen the bug immediately, despite my lack of artistic ability.
At #GoogleIO, we got a private test of Project Astra (not sure I was allowed to film this?). @skirano talking to Astra while it identified what it was looking at. Full multimodal, no lag, super smooth. Insane. pic.twitter.com/ayWPZRiEYq
— Conor Grennan (@conorgrennan) May 15, 2024
No, you were not supposed to record. But here’s a similar demo posted on X.
To give you a sense of where the technology is today, Google staff didn’t allow recording or photographs in the Astra demo room. They also had Astra running on an Android smartphone, but you couldn’t see the app or hold the phone. The demos were fun, and certainly the tech that made them possible is worth exploring, but Google missed an opportunity to showcase how its AI technology will impact your everyday life.
When are you going to need to ask an AI to come up with a band name based on an image of your dog and a stuffed tiger, for example? Do you really need an AI to help you find your glasses? (These were other Astra demos from the keynote.)
Image Credits: Google demo video (opens in a new window)
This is hardly the first time we’ve watched a technology event filled with demos of an advanced future without real-world applications or those that pitch conveniences as more significant upgrades. Google, for instance, has teased its AR glasses in previous years, too. (It even parachuted skydivers into I/O wearing Google Glass, a project built over a decade ago, that has since been killed off.)
After watching I/O, it feels like Google sees AI as just another means to generate additional revenue: Pay for Google One AI Premium if you want its product upgrades. Perhaps, then, Google won’t make the first huge consumer AI breakthrough. Like OpenAI’s CEO Sam Altman recently mused, the original idea for OpenAI was to develop the technology and “create all sorts of benefits for the world.”
“Instead,” he said, “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 in the same boat.
Still, there were times when Google’s Astra AI seemed more promising. If it could correctly identify code or make suggestions on how to improve a system based on a diagram, it’s easier to see how it could be a useful work companion. (Clippy, evolved!)
Gemini in Gmail.Image Credits: Google
There were other moments when the real-world practicality of AI shone through, too. A better search tool for Google Photos, for instance. Plus, having Gemini’s AI in your inbox to summarize emails, draft responses or list action items could help you finally get to inbox zero, or some approximation of that, more quickly. But can it clear out your unwanted but non-spam emails, smartly organize emails into labels, make sure that you never miss an important message and offer an overview of everything in your inbox that you need to take action on as soon as you log in? Can it summarize the most important news from your email newsletters? Not quite. Not yet.
In addition, some of the more complex features, like AI-powered workflows or the receipt organization that was demoed, won’t roll out to Labs until September.
When thinking about how AI will impact the Android ecosystem — Google’s pitch for the developers in attendance — there was a sense that even Google can’t yet make the case that AI will help Android woo users away from Apple’s ecosystem. “When is the best time to switch from iPhone to Android?”, we posed to Googlers of varying ranks. “This fall” was the general response. In other words, Google’s fall hardware event, which should coincide with Apple’s embrace of RCS, an upgrade to SMS that will make Android messaging more competitive with iMessage.
Simply put, consumers’ adoption of AI in personal computing devices may require new hardware developments — maybe AR glasses? a smarter smartwatch? Gemini-powered Pixel Buds? — but Google isn’t yet ready to reveal its hardware updates or even tease them. And, as we’ve seen already, with the Ai Pin and Rabbit’s underwhelming launches, hardware is still hard.
Image Credits: Google
Though much can be done today with Google’s AI technology on Android devices, Google’s accessories like the Pixel Watch and the system that powers it, WearOS, were largely overlooked at I/O, beyond some minor performance improvements. Its Pixel Buds earbuds didn’t even get a shout-out. In Apple’s world, these accessories help lock users into its ecosystem, and could someday connect them with an AI-powered Siri. They are critical pieces to its overall strategy, not optional add-ons.
Meanwhile, there’s a sense of waiting for the other shoe to drop: that is, Apple’s WWDC. The tech giant’s Worldwide Developer Conference promises to unveil Apple’s own AI agenda, perhaps through a partnership with OpenAI…or even Google. Will it be competitive? How can it be if the AI can’t deeply integrate into the OS, the way Gemini can on Android? The world is waiting for Apple’s response.
With a fall hardware event, Google has time to review Apple’s launches and then attempt to craft its own AI moment that’s as powerful, and as immediately understandable, as Steve Jobs’ introduction of the iPhone: “An iPod, a phone, and an Internet communicator. An iPod, a phone… are you getting it?”
People got it. But when will they get Google’s AI in the same way? Not from this I/O, at least.
We’re launching an AI newsletter! Sign up here to start receiving it in your inboxes on June 5.
