Bing AI and the Dawn of the Post-Search Internet
About a year ago, I wrote a column about growing user frustration with Google Search, as automated summaries, sponsored content, and SEO-friendly spam increasingly took away the useful website results that Googling was supposed to generate. Google’s search algorithm wasn’t directing us to the information we wanted to find (like, in my case, the then-inevitable perfect toaster), it was bombarding us with half-baked suggestions from content factories. However, Google Search has maintained its dominance partly out of habit and partly because no competing service has yet offered a viable alternative. On February 7, Microsoft began beta-launching a version of its search engine, Bing, in the form of an AI chatbot powered by GPT-4, the latest iteration of OpenAI’s ChatGPT multilingual model. Instead of directing users to external sites, the new Bing can simply generate its own answers to any query. Google sees the tool as an existential threat to its core business for good reason. At the end of last year Times reports that the company has announced a “red code”. Microsoft’s vice president of design, Liz Danzico, who helped develop the Bing AI interface, recently told me: “We are in the process of searching.”
The Bing AI I’ve had the chance to test in recent weeks is essentially ChatGPT tied to Microsoft’s search directory. Using it is like talking to a very powerful librarian whose authority spans the breadth of the Internet. The old search experience is almost second nature to today’s internet users. type relevant keywords into Google, press Enter, and sift through the list of links that appear on the results page. Then click to find the information you’re looking for. If you don’t find it, maybe go back to the Google Search page and adjust your keywords to try again. With Bing AI, websites are the source material, not the destination, and the results are generated through what Danzico called a “co-creation process” between the user and the bot. Bing AI summarizes digests and aggregates aggregators, sifting through the barrage of online information for you. I asked what toaster Wirecutter recommended and it told me the Cuisinart CPT-122 2-Slice Compact Plastic Toaster; I then asked her to make a list of other recommendations, and she pulled them together from outlets including The Kitchn, Forbes:, and The Spruce Eats. Within seconds and without leaving the Bing AI page, I had a summary of reputable devices. However, the chatbot wouldn’t tell me exactly which one it bought. “I can’t make decisions for you because I’m not human,” it says.
In some ways, a Bing AI user has more agency than a Google Search user. Interacting with a chatbot involves what Microsoft’s senior product marketing manager, Sarah Mody, described as “learning not to speak the quest of twenty years ago.” You communicate in simple sentences rather than single keywords, and you can narrow or refine your query results by asking the machine to follow you. If you ask for directions for a trip to Iceland, for example, then ask: “Where does the sun go down there?” the bot will understand which “there” you are referring to. But in other ways, the Bing user is more limited and passive, encouraged to let the machine decide what information is valuable rather than doing the search on its own. “It’s your travel guide, your bank, your confidant, your guide,” Danzico said. The Chat Mode interface, which is a single chat box above a subtle color gradient, reflects the purpose of that single window. Where using Google Search is sometimes like designing the right equation to solve a problem, using Bing AI is a bit like a series of text message conversations. It even punctuates answers with a smiling, blushing emoji; “I’m always happy to talk to you. 😊” I was told. To the left of the chat box is a button that reads “new thread” and shows the broom sweeping the dust. Clicking it deletes the current conversation and starts over. The module, Danzico told me, was developed with the help of AI itself.
While tools like Bing AI promise extreme, almost unimaginable convenience for users, they will likely be worse for content creators than the search and social media companies that have grabbed most of digital advertising dollars over the past decade. Bing AI suggests links to websites in the form of annotations that link to URLs. But the URLs are deliberately obtuse to minimize what one Microsoft employee described to me as the “cognitive burden” of clicking and scrolling through links. In a video interview the other day, Modi showed how he can ask Bing AI to find a good vegetarian recipe for dinner. The bot pulled it up Bon Appetit Vegetarian Lasagna Recipe (How To) The New Yorker, Bon Appetit owned by Condé Nast) and reprinted it in full in the chat. Modi then asked her to list all the ingredients and arrange them down the aisle of the grocery store, a request no cooking website could hope to fulfill.
Later, on my own, I asked Bing AI to tell me the latest news about First Republic Bank and the larger unfolding banking crisis. It created NBC, CNN and CNN’s breaking news roundup, articles Wall Street Journal, which is paid. (The: Magazine said that any AI referencing its material should pay for the appropriate license, although it might be difficult to apply this to publicly available articles, since AI search crawls the entire web just like Google does.) I then asked Bing to present the news in the style of an Axios newsletter publication. The result was a somewhat dry but otherwise convincing imitation. Another time when I asked Bing for wallpaper options suitable for bathrooms with showers, it returned a list of manufacturers. Instead of looking for a list, I would “co-author” with the bot.
Much of the current web is built around aggregation: product recommendation lists on The Strategist, movie review summaries on Rotten Tomatoes, restaurant reviews on Yelp. What value will those sites have when AI can do the aggregation for us? If Google Search is an imperfect book index that tells us where to find the material we need, Bing AI is SparkNotes, allowing us to bypass the source altogether. Users can simply “read” publications in the form of AI chat summaries, as if they were listening to a mechanized attendant speaking aloud newspaper headlines. However, the paradox of artificial intelligence is that it relies on source material—a vast sea of information generated by other websites—to generate its answers. Because of this, it’s easy to imagine a kind of vicious circle driven by the widespread adoption of tools like Bing AI, where if users no longer have to visit websites directly, those websites’ business models based on advertising and subscriptions will collapse. . But if those sites can no longer produce content, AI tools won’t have fresh, reliable material to digest and recover.
Speaking of fresh and reliable material, what happens when more and more of the material found online is generated by artificial intelligence? Last week, Google and Microsoft both announced a suite of artificial intelligence tools for office workers, with apps that can compose new emails. letters, reports and slide decks or summarize existing ones. Similar tools will almost certainly soon invade other aspects of our digital lives. In New YorkJohn Herman described the likely consequences as “textual hyperinflation”. When a bot is creating content that used to be produced by workers, it can be difficult to determine which parts of your feed’s content make sense; there will be no guarantee that any human work or thought has gone into any given email or report. We are all already bombarded by spam, mostly of the man-made kind. This will be a new type of AI-generated spam on an unprecedented scale that is hard to distinguish from human-generated content. Sooner than we think, content factories can use AI to create full articles, publicists can use it to write press releases, and cooking websites can use it to develop recipes. People will need help navigating captivity, but media companies will have fewer resources to serve that need. In such a scenario, artificial intelligence can solve the very problem it creates. after all, if tools like Bing AI cause the well of online source material to dry up, all that will be left are self-referencing bots digging through the pool. answers that machines were created first.
At the same time, non-automated text online will become an artisanal product, a product we seek for its pristine quality; “natural language” instead of natural wine. On Tuesday, Google announced the launch of its own AI chatbot. Called Bard, a more evocative and lofty title than Microsoft’s dry “AI co-pilot,” it’s a broadside of the AI arms race between the tech giants. But Google has notably separated the chatbot from its own branded product. “We think Bard is complementary to Google Search,” one company executive said said Times. It tacitly acknowledges the company’s current model of the AI threat. In its race to catch up to Microsoft, Google must try to avoid cannibalizing itself. Bing, however, happily leads us into a post-search future. ♦