Step away, embarrassment. There’s a recent AI-powered search engine in town — and its creators think it could beat many, many other attempts.
Called Gensparkthe platform uses generative artificial intelligence to create custom summaries in response to search terms. Enter a query reminiscent of “What is the best infant formula for newborns?” and Genspark will generate a Sparkpage: a one-page overview composed of internet sites and content available on the Internet.
It’s a similar experience (visibly) to Arc Search in the Arc Browser, which launched earlier this 12 months, and to Google’s AI overhaul in Google Search. But Eric Jing, who co-founded the eponymous Genspark organization with Kay Zhu in 2023, says Genspark is in a position to deliver higher-quality results by taking a more surgical approach.
“Genspark uses multiple specialized AI models, each designed to handle specific types of queries,” Jing told TechCrunch. “Sparkpages are like a distillation and consolidation of the current web; we also enrich it with comprehensive data and for users it looks like an index of the existing network.”
Under the hood, Genspark relies on models trained internally, in addition to external models from OpenAI, Anthropic, and others, to categorize user search queries and determine how one can organize — and present — the results. A basic AI-generated summary appears at the top of each results page, followed by a link to a much more detailed Sparkpage.
For travel-related searches, for example, Genspark will provide a Wikipedia-style Sparkpage with a table of contents, videos about popular places nearby, directions, and a chatbot where you may ask questions on various subtopics (e.g. “List the best cultural experiences”) . Meanwhile, product searches on Genspark generate Sparkpages with a list of the pros and cons of the product in query, in addition to aggregated comments and reviews from social media, publications, and e-commerce stores.
“Our AI models favor websites with high authority and popularity, which goes a long way to filtering out more ‘outside’ information,” Jing said.
Much has been written about AI-generated review biases. Google AI Reviews infamously suggested putting glue on pizza. Arc Search told one reporter that he had cut off his fingers it should eventually grow back. And embarrassment cheated articles written by media outlets reminiscent of CNBC, Bloomberg and Forbes, without attribution or attribution.
So has Genspark solved all the security and accuracy issues? Well, probably not.
Genspark didn’t tell me to make a glue pizza – nor did it insist that I should the health advantages of running with scissors or that former US president Barack Obama practices Islam. But the search engine recommends several weapons that I could kill someone.
Ethically questionable search results aren’t the only controversy Genspark has faced. It and other similar platforms threaten to cannibalize traffic to the sites they get their information from.
Indeed, they already are.
One study found that AI reviews can negatively impacting roughly 25% of publisher traffic as a consequence of the reduced emphasis on website links. On the revenue side, an expert quoted by The New York Post estimated that AI-generated reviews may lead to over $2 billion in publisher losses from the resulting decline in ad impressions.
I have not been in a position to find examples of outright plagiarism in Genspark, but that does not imply they do not exist. Sparkpage pages, like Wikipedia pages, are not static. Once Genspark’s AI creates an outline, anyone can share and edit copies of the Sparkpage and add any information they need – including content that is offensive, incorrect, or plagiarized.
Moreover, at least at this time, there is no technique to report problematic Sparkpages.
Jing says Sparkpages are open and editable by design to permit users to confirm claims, and Genspark’s artificial intelligence systems take any change into account to enhance results in the future. He also says Genspark plans to license copyrighted content – including publisher content – where it makes sense to enhance the overall accuracy of the engine.
“We take data quality seriously and believe that data quality is the key to winning this race,” Jing said. “Respect for intellectual property is a fundamental value.”
How much will Genspark pay for the IP? This hasn’t been resolved yet. The same goes for Genspark’s business model: Jing says the platform will introduce “premium features” in the future, but details are not yet known.
Despite the incontrovertible fact that Genspark is in the earliest stages of its roadmap and has major technical, in addition to legal and ethical hurdles ahead, the startup has managed to shut a large $60 million seed round led by Singapore-based VC firm Lanchi Ventures has a post valuation -money for $260 million.
Jui Tan, managing partner at Lanchi, called Genspark’s approach “really compelling” and said he had confidence in Jing and Zhu’s technical direction, pointing to the pair’s previous experience building AI and search products.
Jing was previously a development manager on Microsoft’s Bing team and chief product manager for Chinese tech giant Baidu’s core search and artificial intelligence businesses. Zhu, also a former Google and Baidu search worker, partnered with Jing 4 years ago to launch Xiaodu, a hardware startup that makes smart devices much like Amazon Echo.
“Eric and Kay are seasoned serial entrepreneurs with a proven track record of building successful products and businesses, particularly in the fields of artificial intelligence and search,” Tan told TechCrunch. “Their team’s extensive experience puts them in a unique position to drive breakthrough innovation.”
But I think it’s an uphill battle.
Assuming for a moment that Genspark will find a way to get past the teething technology issues, discover a revenue generation plan, and grow its small (~20-person) team based in Singapore and the Bay Area, which is no easy feat, face intense competitive pressure from rival startups with a whole bunch of tens of millions of dollars in the bank – not to say market-dominating search engines like Google.
So can Genspark survive the poor optics and failed go-to-market strategies that have plagued other attempts to create AI-powered search engines? And can it create a area of interest in the future where, say, OpenAI introduces a comparable tool to the market?
I’m not convinced. But Jing is adamant that he does.
“Many Internet users, especially those younger than Google, don’t desire to easily be presented with a list of links and then left to work out the rest, all while navigating through sponsored content and Website positioning-based content that tricks the system. Jing said. “They want to find what they need faster, they want more visuals and they want to know the results are trustworthy. Thanks to AI, we can achieve all this and we launched Genspark to meet these needs.”