MOUNTAIN VIEW, Calif. — Has AI found its VisiCalc?
That’s what I’m wondering after spending Wednesday at Google I/O, the search giant’s annual developer conference at Shoreline Amphitheatre. Have we seen anything that will become to artificial intelligence what spreadsheets were to personal computers? Something so amazing and useful that it compels people to embrace a whole new form of technology?
It’s probably too early to tell. But my hunch is that we’ve seen at least one or two potential components of what could become AI’s killer app* in what’s been shown and released so far by companies like OpenAI, Microsoft, and Google.
Along those lines, three projects stood out to me at Google I/O.
- Project Tailwind: This AI notebook uses your notes and source materials to create a “personalized and private AI model” that answers questions and summarizes content. It was pitched for students, but this approach could transform my job, and many others. Described as an experimental project by five Google engineers, it’s an example of the ideas bubbling up organically in AI right now. There’s a waitlist to try it.
- Google Search AI: For its flagship product, the company showed the potential for generative AI to supplement search results by summarizing key points related to a search. One example on stage was the search, “good bike for a 5-mile commute with hills,” which returned an AI-generated snapshot of tips to keep in mind, in addition to traditional search results and reviews.
- Help Me Write: This upcoming feature for Gmail in Google Workspace drafts entire messages for users. An example at I/O was a message seeking a refund for a canceled flight, with the option to generate a more elaborate version, under the theory that it would boost the chances of success.
Addressing the crowd during the I/O keynote, Google CEO Sundar Pichai pointed out that Help Me Write is an evolution of Gmail’s Smart Reply (quick suggested responses) and Smart Compose (suggested auto-complete for sentences).
This was his way of underscoring the fact that Google has been into generative AI for years, even if the pace has been accelerated in recent months by ChatGPT maker OpenAI, whose partnership with Microsoft has supercharged the Redmond company’s progress in these areas.
In other words, Google’s Bard chatbot is just part of the picture.
“We have been applying AI to make our products radically more helpful for a while,” Pichai said on stage at Google I/O. “With generative AI, we are taking the next step.”
There were plenty of other interesting AI projects at Google I/O. In fact, there was a full 80 minutes of AI before they even got around to showing the Pixel Tablet with smart dock (a new rival to Amazon’s Echo Show) and the ($1,799!) Pixel Fold smartphone (notable in part because of Google’s collaboration with Microsoft on the Android-powered Surface Duo dual-screen devices).
Back to the topic at hand, a couple other AI projects also stood out:
- Magic Editor: An example of generative AI’s potential in photo editing, this Google Photos feature will let users alter and add to pictures. One example shown at I/O was a photo of a young kid holding balloons on a bench, where the AI recentered the photo and synthetically added to the bench and balloons, to make that recentering possible.
- Watermarking and metadata: As part of the ongoing societal discussion about AI ethics and corporate responsibility, Pichai said Google will ensure that images generated by AI include indications on the images and in the underlying metadata identifying them as synthetic, and providing more information about how they were generated.
But for my purposes, the generative AI Google search summary, Project Tailwind, and Help Me Write seem to be headed in the direction of something game-changing.
Personally, my killer app would start with a true, AI personal assistant to read, summarize, and prioritize my email. I spoke at the event with Aparna Pappu, the Google Workspace vice president, who assured me that the engineering team is sympathetic to and interested in this challenge.
As a bonus, or Version 2, it would be great for this hypothetical app to also include an AI summary of my social media, text messages, etc., not just my email.
Other companies are also headed down this path:
- As part of its Slack GPT announcement, Slack recently showed a new AI feature for summarizing conversations and content inside the collaboration app.
- Microsoft offers AI-generated summaries based on a company’s internal content within Microsoft 365 Copilot, its enterprise AI tool now in private testing. Microsoft’s Build developer conference, its counterpart to Google I/O, is coming up in a couple weeks.
- Premium email app Superhuman seems to developing components of this killer app, as well, with its generative AI features, plus email prioritization.
- I’m also a user and fan of the premium service SaneBox for prioritizing emails.
- There are also existing capabilities for message detection and summarization in the intelligent features of Google Assistant, Apple Siri, and Amazon Alexa.
Lastly, I’d like this hypothetical app to prioritize these messages in a simple to-do list, with suggested replies and/or recommended actions based on its learned understanding of my preferences and communication style.
Does this all-encompassing AI personal assistant already exist? I haven’t found it yet. Does the idea resonate with you? What would be your AI killer app? Let me know.
* I asked Bard, ChatGPT, and Bing the same question: “What are some examples of killer apps in technology’s evolution?” See and assess the results for yourself.
More notes from Google I/O …
The new version of Google’s Project Starline high-fidelity telepresence system is jaw-dropping. I got a short demo at the event, and it was the closest I’ve ever experienced to being in the same room with someone who isn’t there. Cost remains the primary hurdle to getting this out into the real world. See Google’s video above.
Google Cloud showed a new capability called Duet AI that generates code recommendations and other forms of assistance for cloud developers. Duet AI joins a field that includes Microsoft’s GitHub CoPilot and Amazon’s Code Whisperer, and offers an indication of where Google’s cloud unit is headed in the realm of AI.
Composer and musician Dan Deacon set a new bar for pre-show musical entertainment and enlightenment with his AI-assisted songs, live vocals and comedic nerdery, leading a guided meditation culminating in a giant duck with lips appearing alongside him in real life. It was great. Watch the video here.
The teleprompter went out at one point during Pichai’s keynote address, according to the behind-the-scenes scuttlebutt afterward, but the Google CEO was apparently able to carry on from memory, using the slides as his cues.
A plane buzzed Shoreline Amphitheatre repeatedly during the first half of the keynote, pulling a banner calling on the company to protect abortion privacy.
Watching the demos, particularly those focused on content generation and photo manipulation, my mind kept going back to this question: What happens to the fulfillment, satisfaction, and joy that comes from human creativity when machines are better at creating than we are?
Now back to manually sifting through the 332 “important and unread” messages in my Gmail inbox after a few days of travel for in-person events. I’m not bragging; I’m complaining. That killer AI app can’t get here soon enough.
