As the ChatGPT hype hits fever pitch, Neva launches its generative AI search engine globally. • TechCrunch


Challenger search engine Neva wants to replace the usual “10 blue links” in search results with something more suitable for the modern AI age.

In December, Neva founder and CEO Sridhar Ramaswamy, who previously led Google’s ad technology business, scoffed. brand new He likens himself to ChatGPT’s hype train of “cutting edge AI” and large linguistic models (LLMs).

“ChatGPT cannot provide you with real-time information or fact-checking,” Ramaswamy wrote. “With our latest upgrades, Neva can.”

Fast forward to January, and Neva has officially launched NevaAI into the US market, billed as “real, real-time AI search”. While it’s technically possible to access NeevaAI for users around the world, incorporating language and locale preferences into account settings required a bit of grouse-pokery. But today, NeevaAI is officially rolling out globally to registered users including Canada, UK, Germany, France and Spain. In addition, the Neeva search engine (not NeevaAI) is moving to Australia and New Zealand.

The timing of today’s announcement comes especially a week after Microsoft reignited the search engine wars by introducing ChatGipt to Bing’s search engine, and with it the promise of a completely revamped search experience.

What is ChatGPT?

Most years can be defined by at least one overwhelming technology trend. In the year In 2022, Web3 was one of the biggest buzzwords in town, with metaverse and tangentially immersive technologies clamoring for mindshare. Although there is nothing to show that such trends will dry up in the near future, it is clear from the first six weeks of the new year that creative AI will be the big talking point of 2023.

Generative AI, basically, describes the process of using algorithms to create (“generate”) new content. The poster child of this movement is ChatGPT, a chatbot-style technology trained by large linguistic models (LLMs) that can do very nice (but flawless) things like essays, articles, poetry, poetry, and computer programs. The OpenAI prototype, part of Microsoft-backed artificial intelligence (AI) research, has taken the world by storm since its launch in early November, with some arguing that open AI has reached the mainstream.

Touting its OpenAI service as a premium subscription, Google last week unveiled a new “experimental conversational AI service” called Bard, which aims to provide more insight into how AI is changing search engines by leveraging integrated responses from multiple sources. Correct answers to online questions. And that’s what Microsoft launched two days later – a new incarnation of Bing powered by an advanced version of ChatGPT, optimized for real-time search.

Instead of serving up the usual list of ladder links, the new AI-infused search engine takes a question, scans the answers and provides a response full of citations to the original sources.

The new AI-infused Bing search engine is in action. Image credits: Microsoft

And this, effectively, Neva is now bringing it to international markets, just a month after its US launch.

The story so far

Briefly, Mountain View, Calif.-based Neva first launched the subscription-only search engine in the domestic US market in June 2021 and later added a free “basic” tier with some restrictions. place. The company brought the search engine to Europe in October, and has since released it to additional markets around the world.

Neva’s main selling point is that it doesn’t monetize ads and prevents third-party trackers from using personal data to show you personalized ads – Neva wants to make money through good old-fashioned paid subscriptions.

An ad-free search environment means users don’t have to scroll through countless sponsored results to get to the organic links they want. However, the basic look and feel of Neva is the same as search engines have been since before Google arrived on the scene – with a few aesthetic differences thrown into the mix after a series of individual sources.

Neva: Only organic search results Image credits: Neva

But with NeevaAI, the startup is looking to do its part by reinventing search.

TechCrunch has been messing around with NeevaAI for the past few weeks, and from our tests it’s doing pretty impressively and comfortably handling questions like “Why the Beatles were so big.? or “What is the world 5k [running] Record? ” Neva is generating real-time feedback from multiple cited sources.

These quotes are key to avoiding the “black box” controversies that surround many AI technologies. The idea here is not only to promote transparency by showing people where to get the information, but also to increase the likelihood of continued referral traffic by giving the content they publish on the website the credit they deserve.

NeevaAI in action. Image credits: TechCrunch

But some questions are simply not ready to be tackled. In those cases, it defaults to known website-specific links and snippets so users can see and discover the answer themselves.

Neva is returning normal results. Image credits: TechCrunch

In an interview with TechCrunch for this story, Ramaswamy explained some of the logic that went into developing NevaAI, including deciding how and when to generate answers from multiple sources.

“Essentially speaking, it works for questions so we can find authoritative websites that answer the question,” Ramaswamy explained.

So, unlike the many amazing examples we’ve seen so far where ChatGPT is tasked with writing lyrics in the style of Nick Cave, for example, NeevaAI won’t help you here. Likewise, if you ask her a trick question, intentionally or otherwise, she may not answer, refer back to the usual list of links to check for yourself.

“Our goal with NevaAI is – first and foremost – to be safe, we don’t want to tell you the wrong things,” continued Ramaswamy. “So we’ve chosen to be safe in how we bring it (information) in.”

An example here would be if someone asked NevaI when Boris Johnson served as King of the United Kingdom, rather than the actual question of when he served as Prime Minister. That is the question. can It can fool any search engine, because there will undoubtedly be pages containing all the words in the query – so it’s up to Nevai to make sure it understands the query and respond appropriately. Or there is no response, which is what he does when he does not trust the answers to questions.

NeevaAI declined to answer questions. Image credits: TechCrunch

Obviously, NeevaAI’s response here is not perfect. A more intelligent response – as one suggests – is to tell the user that Boris Johnson is not actually the King, but he has served as Prime Minister. Or at the very least, provide a more satisfactory response for the user to clarify the question. And that’s something that could come to NeevaAI in a future iteration, though it’s unclear what form this will ultimately take.

“If you ask some stupid question, it doesn’t mean anything because there’s nothing on the web to indicate whether it’s true or false,” Ramaswamy said. “So if we don’t have an answer, we refrain from giving an answer. But we will fix it soon.

Game mode

TechCrunch’s chat with Ramaswamy came before Bing’s big announcement last week, so today’s global market expansion has a slightly different feel to Microsoft doing essentially the same thing — billions of dollars more to help the box. And it’s now clear that Google will do the same in the future.

Elsewhere, other search engine startups are following a similar trajectory to Neva. In December, venture-backed launched ChatGPT-style AI smarts similar to Neva.

Neva clearly has its work cut out for it in a market that includes long-standing billion-dollar incumbents and other agile startups. However, at a time when the world is weary of big tech’s use of big data, Neva hopes its ad-free approach will win over its fans to further flourish as a search engine. The likes of doesn’t serve ads yet, but says it may offer “non-privacy-invasive” ads in the future.

Above all, Neva is betting on its future by developing its own search stack, rather than relying on the same engine used by Google or Microsoft Bing as other challengers.

None of this is to say that Neva is a completely independent, autonomous entity. While it is pushing to fully rely on its own search stack, for now it relies on Bing some Web search activity. Separate to NeevaAI, Ramaswamy confirmed that it works with “larger language modeling companies,” including OpenAI’s GPT-3.5 and Cloud, the high-venture-backed Anthroposic founded by former OpenAI employees.

“We use them for things like generating training data for us, and sometimes we make calls for them to help with summaries,” Ramaswamy said. “Most of the models we use for NeevaAI are our own – we pre-train or adapt them for custom tasks like question-answering or summarization.”

Founded four years ago, Neva has received $77.5 million in venture capital funding from Silicon Valley VCs including Sequoia Capital and Greylock. So while it’s a contrast to the likes of Google and Microsoft, among other well-funded startups, they’re not without friends and funding.

Regardless of where Neva goes from here, it’s clear that 2023 is shaping up to be a big year for generative AI, with the likes of Shutterstock recently opening up to create stock images based on text queries.

Similarly, Neva has no plans to limit the scope of its AI efforts to simple search engine-based data retrieval and is likely to expand into related verticals in the future.

“Vivek [Neeva co-founder Vivek Raghunathan] And I’m like kids in a candy store, we don’t know what to do next,” Ramaswamy joked. “There’s so much—simple, natural tracking stuff. Generative AI is one direction, site-specific search is another extremely interesting direction. They are very close to what we do.”

NeevaAI is available today in both free and premium plans for English-language users in the UK and Canada, as well as local language incarnations in Germany, France and Spain.


Source link

Leave a Reply

Your email address will not be published. Required fields are marked *

8 + 13 =