The long-planned AI chatbot war is officially on. And, as tech companies rush to modernize their services, experts warn (opens in a new tab) on the dangers of rushing this ChatGPT frenzy into research.
Ad-free search engine Neeva, however, promises to fix AI-based search by combining the power of generative AI systems with its authoritative and unbiased search technology.
This means that, unlike ChatGPT apps, its results are real-time. Additionally, its AI summaries are backed up with source links so users can self-check their reliability.
First launched in December 2022 in the US, NeevaAI beat Big Tech giants Microsoft and Google by developing its own ChatGPT-like search assistant. Microsoft Edge’s ChatGPT tool was actually announced just a week ago, to complement Bing’s AI search. So was Google’s “experimental conversational AI service” Bard.
And now, NeevaAI has now expanded its reach outside of the United States. Starting February 13, 2023, users living in Canada, the UK, Germany, France and Spain will be able to experience the power of its AI-powered search results.
But can NeevaAI Really solve the riddle of AI-powered search?
✨ You asked. We delivered. ✨ Today NeevaAI is officially launched and ready to try in:🇨🇦 Canada🇫🇷 France🇩🇪 Germany🇪🇸 Spain🇬🇧 United Kingdom Just login to your Neeva account and start searching at https://t .co/aZwoDBErAs pic.twitter.com/iGhxn00hs5February 13, 2023
From a search engine to an answer engine
“Search, which is a daily use function, has become more and more in the service of advertisers and not in the service of users. Neeva turns the whole model on its head and creates an AI-powered search, ad-free, completely designed for users,” said the CEO of Neeva. Sridhar Ramaswamy told TechRadar.
Bringing back a user experience first was, in fact, the main reason Ramaswamy left his position as Google Ads Manager to build his private ad-free search engine in 2018. Today, with the explosion of a software like ChatGPT, he sees a new opportunity to better serve Neeva’s customers.
The algorithms that power generative AI software, like ChatGPT, use large language models (LLMs) to create new content. NeevaAI (opens in a new tab) combines this innovative capability with the authority and speed of its software to transform “a search engine into an answer engine”.
Such a system is built to query a page, understand its content and decide whether or not it might be useful and/or authoritative enough to be cited as a source. It does this in real time as the web changes, to provide a single synthesized answer with linked sources gathered from the sites most relevant to a query.
Ramaswamy explained that NeevaAI primarily uses its own models that its team can pre-train or refine for custom tasks such as question answering or summarizing.
These are integrated with other systems, like GPT3.5 and Anthropic’s Claude, to generate training data (used to train its own models) as well as super-specialized tasks in the overall pipeline.
While this new accessibility of generative AI systems is one of the biggest technological innovations of recent years, the early iterations are far from perfect.
Software like ChatGPT actually has two significant drawbacks. First of all, its results are never sourced or referenced. Second, these are not fetched for a real-time data stack. The results will then be less relevant and more difficult to verify.
While some other big tech companies are launching an AI-powered search chatbot to try to address this last point, Neeva’s unique feature provides citation cards linked to its results. These swipeable cards highlight authoritative information on the topic being researched, while suggesting important research questions for the researcher.
“Our goal has been to integrate AI responsibly and provide authoritative answers that users can trust.”
Ramaswamy thinks Neeva’s ad-free model can also be beneficial here.
He said: “Every time you move a product into a mode of asking a question and giving a single answer – the ads fall apart. What we’re seeing is the tension in both achieving the full potential of this technology and the fact that these companies have significant revenue protection.
“Not only that, but we were able to deliver this six weeks before the big techs announced their own releases (which are still not available yet).”
After receiving positive feedback after its launch in December, the company has now launched its AI search assistant also in Canada and a few European countries. However, people living outside of these countries could also enjoy the feature just by playing around with some settings.
Additionally, Neeva is so far the only one in the game to commit to supporting publishers and content creators – those who are suffering the most from the rise of AI chatbots. In addition to offering them 20% of its revenue when their content is used to answer a query directly, Neeva is also exploring ways to help publishers leverage AI on their sites for a better user experience.
AI chatbot race: Is NeevaAI really that good?
But, while it’s true that Neeva likely started the race for AI chatbots, she’s certainly not the only player in the game today.
Among the biggest names, Microsoft has just launched its new AI-powered Bing based on the OpenAI system. Google responded with Bard. The latter uses Google Lamda’s own language learning model (opens in a new tab)which was considered “sensitive” by one of the engineers who worked on it.
Smaller browser You.com has also integrated its site with a similar AI chatbot, and China-based search engine Baidu is rolling out its Ernie in March. Outside of the search world, tech giant Meta already launched Blenderbot last summer in the US.
Tech companies seem to be rushing to put their fingers in the AI pie. However, the results we are seeing so far are far from perfect.
Indeed, ChatGPT-type software has certain intrinsic limitations.
For example, something known as jailbreaking can allow the AI chatbot to develop dangerous and abusive ways of responding to users. But, perhaps most problematic is the tendency to sell misinformation as fact. This issue has already cost Google more than $100 billion because Bard responded to a query with false information at its launch event.
My new favorite thing – Bing’s new ChatGPT bot argues with a user, informs them that the current year is 2022, says their phone might have a virus, and says “You haven’t been a good user” For what ? Because the person asked where Avatar 2 shows nearby pic.twitter.com/X32vopXxQGFebruary 13, 2023
On this point, Ramaswamy said that Neeva works differently than pure generative AI because it uses a more constrained system that will refuse to respond in case it cannot find an appropriate answer.
However, this does not mean that Neeva will never report misleading or false information.
For example, when we asked NeevaAI the same question that tripped up Google’s Bard at its launch event – “What new discoveries from the James Webb Space Telescope can I tell my 9-year-old about?” – we were presented with the exact same answer that Bard gave.
This appears to be because NeevaAI found valid sources reporting the news, but was unable to discern that those sites were actually reporting that this answer was wrong.
“That’s why quotes are central to Neeva’s answers on AI,” Ramaswamy said. “Just as no website is credible in its entirety, reflection and skepticism are key to surviving in this modern world.”
