The first half of February was huge for the future of internet search. But it wasn't a perfect launch.
Generative AI is having another huge, headline-generating month, which is becoming a regular thing at this point. February kicked off with Google and Microsoft elbowing each other in and out of the spotlight with news of upcoming releases that will bake generative AI into their respective search functionalities.
The product previews as a whole generated a lot of buzz around the future of web searches, search-based advertising models, SEO and other topics. But after those previews were dissected, it was discovered that the prompted outputs shown in demonstrations from both companies had errors presented as facts.
Overall, the updates set the stage for our interactions with the internet to fundamentally change — but not quite yet. Let’s dissect all the good and cumbersome generative AI-search news from Google and Microsoft in the first half of February.
Monday, February 6: Google announced via a blog post that it will be releasing Bard, a conversational AI service built on top of its Language Model for Dialogue Application (LaMDA), to the public in the coming weeks. Bard is meant to be a direct response to OpenAI’s Chat GPT and CEO Sundar Pichai said in the post that more information about its LLM-based evolution to search would be given at a press conference in Paris on the 8th.
Tuesday, February 7: Microsoft made a surprise jump into the ring by announcing via video that a more advanced version of GPT 3.5, the engine that powers Chat GPT, is being used to enhance internet searches and user experiences on Bing and Microsoft Edge. The new GPT is labeled as the “Prometheus Model:” it improves the validity of search queries by offering up-to-date details rather than having its knowledge capped at a certain timestamp (before 2022) like its predecessor. The new functionality was immediately available for limited use on Bing and Edge, however, the full version is available via waitlist and was said to be accessible to the general public in the next few weeks.
Wednesday, February 8: Google hosted its widely-anticipated conference in Paris. The hype for the event was at a fever pitch based on Microsoft’s move the day before. The Alphabet company showed very brief, gif-based snippets of Bard in action, as well as upcoming developments to Google Maps, Translate and Google Arts & Culture all driven by generative AI.
Also Wednesday, February 8: Google’s gif demonstration of Bard’s capabilities showed the AI making a factual error in its output to the prompt: "What new discoveries from the James Webb Space Telescope can I tell my 9 year old about?" The output said the James Webb telescope took the first image of an exoplanet (a planet outside of our solar system) when it was actually taken by European Southern Observatory's Very Large Telescope. That inaccuracy cost the search titan $100 billion in market value after its shares dropped around 9 percent for the day, according to Reuters.
Monday, February 13: Eagle-eyed search engine researcher Dmitri Brereton investigated the accuracy of the results presented during the new Bing’s announcement on the 7th. He discovered that the search engine presented a wide variety of factual inaccuracies and hallucinations much like the issue found in Bard’s presentation (but Bing’s errors were greater in number).
The tech world was waiting for Google to respond to Chat GPT after it was discovered that GPT caused a “code red” at the Alphabet company. It was reported that Google’s CEO Sundar Pichai immediately diverted personnel and resources away from other projects to focus more on implementing AI into its products. Bard’s initial announcement offered a glimpse into the results of those emergency efforts (and what Google has been doing with LaMDA in recent years).
“Soon, you’ll see AI-powered features in Search that distill complex information and multiple perspectives into easy-to-digest formats, so you can quickly understand the big picture and learn more from the web: whether that’s seeking out additional perspectives, like blogs from people who play both piano and guitar, or going deeper on a related topic, like steps to get started as a beginner,” Pichai said in Bard’s announcement post.
Bard’s news on February 6, plus the scheduling of a press conference to learn more about the release, created substantial excitement in the tech news cycle. But it was Microsoft’s move the following day that brought the ongoing chatter about the future of generative AI and the internet to a roar. Microsoft’s decision to both announce and release its Bing and Edge updates at that moment were strategic. The company had been silently taking advantage of its $10 billion OpenAI investment and was waiting for the right moment to show the world those developments.
“I think this technology is going to reshape pretty much every software category,” Microsoft CEO Satya Nadella said in a press conference at the company’s Seattle headquarters on February 7. “The race starts today, and we’re going to move and move fast. Most importantly, we want to have a lot of fun innovating again in search, because it’s high time.”
Before anyone got a sense of just what Bard could really do (since it was/is still being tested behind closed doors), people like Tom Warren, senior editor at The Verge, and thousands of others were already toying with the new and improved Bing and Edge.
“It’s pretty bold of Microsoft to open this up to the public inside its own search engine and let people play around with it freely,” said Warren, after spending some time testing the new search capabilities.
“I asked it what Microsoft had announced during its Bing AI event, and it produced a summary of the news based on multiple publications covering Microsoft’s AI-powered Bing announcement. That was just minutes after Microsoft had made the news official.”
Warren went on to discuss some of the ways that Microsoft is capitalizing on OpenAI’s tech by both implementing and evolving Chat GPT’s functionality, specifically its chatbox. He said there’s a sidebar chatbot that users can run queries in based on individual web pages they’re looking at.
“Imagine you’re on an Amazon listing about a TV — you could ask Bing to compare the best 4K TVs, find the best price and then formulate this all into a neat little table that you can email to your partner,” he said.
While Warren and many others were testing Bing and Edge (even more on that shortly), the Google team was in Paris delivering news about their AI-infused search functionality. Pichai showed off Bard’s NORA capabilities (“no one right answer”) via gif. The implementation will present more nuanced responses to complex or subjective queries in conversational language. However, the Bard demo was brief and a release date wasn’t offered, nor any publicly testable version at the time of the press conference.
Then came the error from outer space.
As mentioned above, Google’s gif of Bard in action contained an incorrect fact about the James Webb Telescope. Once it was discovered, Google’s market value dropped by $100 billion. While the error certainly wasn’t ideal, Wall Street’s response to it was a bit dramatic and many in the tech news cycle positioned Microsoft as the clear winner in this battle as a result.
But that declaration was premature, as evidenced by Dmitri Brereton’s summarization of the errors in Bing’s AI announcement post five days later.
“Bing AI got some answers completely wrong during their demo,” Brereton wrote. “But no one noticed. Instead, everyone jumped on the Bing hype train.”
In the announcement video, Bing presented incorrect information or hallucinations about: nightlife in Mexico City, earnings for Gap and Lululemon, reviews for a pet vacuum and more. Bing’s availability for some public experimentation has also led to some rough results since its release. In one case, a user got Bing to offer “Heil Hitler,” as a suggested prompt. And there are many other instances of the tool being tricked into bypassing its own restrictions or it acting manic (sometimes hilariously so) in response to certain questions or when directed to address its own inaccuracies.
A lot happened in these two weeks — some developments were really exciting while some were a little more awkward. But regardless, both companies are working to bring AI chat interfaces to their already-powerful search engines. All this activity from these household names in tech is great for the future of generative AI as a whole and how the world interacts with the web. We now have glimpses into a future where we get the online insights we’re looking for in a faster and more intuitive way.
The word “transformative” is getting thrown around a lot these days in relation to generative AI. Many are saying that about Microsoft and Google’s investments into gen AI-aided search.
But if the errors in the announcement posts and initial systems tests are any indication, we aren’t there just yet. Microsoft and Google still have a lot of fine-tuning to do before these evolutions in search can be fully relied upon. They’re putting in the time and money necessary to see great outcomes but it’s also vital that they don’t rush things. The world relies on the web and the accuracy of trusted sources to an immeasurable degree. Introducing more avenues for misinformation, bias and unethical content (even if that content is merely a short response from a chatbot) is something the world does not need more of.
Once these systems get some fine-tuning, there will still be questions left to address as the technology grows in adoption. For example:
I think it’s important to remember that the companies building generative AI tools, and everyone using them, are still figuring out the use cases and limitations of this incredibly powerful technology. Alongside that, businesses leaders and marketing professionals have a lot to consider as the nuanced responses to these questions slowly but surely get revealed.
“While we don't know where this is all going to net out, it's important for marketing and business leaders to dedicate the resources to follow its development and adjust strategies for its possible ramifications,” said Jasper’s VP of Marketing Meghan Anderson. “Some of this is just good housekeeping. For example, if you are disproportionately reliant on search or search ads as a major revenue driver, it may be wise to diversify your acquisition channels. Even if search behavior doesn't change, that is best for the health of your website.”
Krista Doyle, Jasper’s search engine marketing specialist, offered her perspective on what the next wave of content marketing may look like within the new AI-driven web.
“Diversification of search channels is something that’s on my mind a lot right now,” said Doyle. “Like content quality, diversifying away from a blog-only SEO strategy was already becoming necessary.
“Not a day goes by now where I don’t come across LinkedIn articles, YouTube video packs and TikTok videos in primary search results. Social networks are now mini search engines in their own right, and as Google and Bing fight to figure out AI’s place in the search engine results pages (and potentially sends less traffic to brands) it will become even more important for teams to redefine organic success and stake their claim in these newer social search channels.”
Collectively, we have to be patient to let the best solutions — those worth trusting the most — come forth. And they will, but with time. This clash of the search titans is only just beginning and the functionality they’re trying to produce will only improve. When it does, we’ll test the solutions, break them, let the updates release, test them again and so forth until they’re sufficiently reliable.
February marked a pivotal moment in the history of the internet. If the success of Chat GPT is any indication, the end users will be the real winners in the Google versus Microsoft battle for AI search dominance. But only once the digital dust settles and these tools are inevitably more trustworthy.
Shopping Jasper for a large business? Update your input to 100+ employees to meet with our team. Please use business email to meet with our team.