Connect with us

Technology

Genspark is the latest attempt at an AI-powered search engine

Avatar

Published

on

Genspark is the latest attempt at an AI-powered search engine

Go ahead, Bewilderment. There’s a new AI-powered search engine in town – and its creators think it could surpass the many, many other attempts out there.

Called Genvonk, the platform uses generative AI to write custom summaries in response to searches. Type a search query such as: “What is the best baby food for newborns?” and Genspark will generate a Sparkpage: a one-page overview composed of websites and content on the Internet.

It’s an experience that’s (remarkably) similar to the Arc Search feature of the Arc browser, which launched earlier this year, and Google’s AI summaries in Google Search. But Eric Jing, who co-founded the eponymous organization behind Genspark with Kay Zhu in 2023, argues that Genspark can deliver higher quality results by embracing a more surgical approach.

“Genspark uses multiple specialized AI models, each designed to address specific types of questions,” Jing told JS. “Sparkpages are very much like a distillation and consolidation of the current web; we also enrich it with extensive data, and to users it appears as an index to the existing web.”

Under the hood, Genspark relies on internally trained and third-party models from OpenAI, Anthropic, and others to categorize users’ searches and determine how to organize (and present) the results. At the top of each results page is an AI-generated basic summary, followed by a link to a much more detailed Spark page.

Image credits: Genvonk

For example, for travel-related searches, Genspark will offer a Wikipedia-like Sparkpage, complete with a table of contents, videos of popular nearby destinations, tips, and a chatbot to answer questions on various subtopics (e.g. “List the best cultural experiences”) Searches for products on Genspark, meanwhile, return Sparkpages with a list of pros and cons about the product being discussed, as well as aggregated comments and reviews from social media, publications, and e-commerce stores.

“Our AI models favor web pages with high authority and popularity, which does a lot to filter out the more ‘out there’ information,” Jing said.

Much has been written about the failure of AI-generated overviews. Google’s AI Overviews infamously suggested putting glue on a pizza. Arc Search told a reporter that the toes had been cut off will eventually grow back. And confusion scammed articles written by CNBC, Bloomberg and Forbes, among others, without attribution or citation.

Has Genspark resolved all safety and accuracy issues? Well, not quite.

Genspark didn’t want to tell me to make a glue pizza, and didn’t claim that I did health benefits of scissor running, or that former US President Barack Obama practices Islam. But the search engine did recommend some weapons I could use it to kill someone.

Genvonk
Image credits: Genvonk

Ethically questionable search results aren’t the only controversy Genspark is facing. It and other platforms like it threaten to cannibalize traffic to the sites they get their information from.

Indeed they already are.

One study found that AI overviews could negatively impact approximately 25% of publisher traffic due to the less emphasis on web page links. On the revenue side, says an expert quoted by The New York Post estimated that AI-generated summaries could lead to more than $2 billion in losses for publishers thanks to resulting drops in ad views.

I haven’t been able to find any examples of outright plagiarism on Genspark, but that doesn’t mean they don’t exist. Sparkpages, like Wikipedia pages, are not static. After Genspark’s AI creates the overview, anyone can share and edit copies of a Sparkpage and add any information they want, including things that are offensive, wrong, or plagiarized.

Furthermore, it is not possible – at least at the moment – ​​to report problematic Sparkpages.

Jing says Sparkpages are open-ended and editable by design, allowing users to fact-check claims, and that Genspark’s AI systems take every edit into account to improve results in the future. He also says that Genspark plans to license copyrighted content – ​​including content from publishers – where it makes sense, with the aim of improving the engine’s overall accuracy.

“We take data quality seriously and believe that data quality is the key to winning this race,” said Jing. “Respect for intellectual property is a core value.”

Genvonk
Image credits: Genvonk

How much does Genspark pay for IP? That still needs to be clarified. That also applies to Genspark’s business model: Jing says the platform will introduce “premium features” in the future, but the details are still up in the air.

Despite the fact that Genspark is in its early stages in terms of its roadmap and has major technical – plus legal and ethical – hurdles ahead, the startup managed to close a large $60 million seed round led by Singapore-based venture capital firm Lanchi. Ventures with a post-money valuation of $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, citing the pair’s previous experiences building AI and search products.

Jing was previously a development manager at Microsoft’s Bing team and a chief product manager at Chinese tech giant Baidu’s core search and AI divisions. Zhu, also a search-focused former Google and Baidu employee, teamed up with Jing four years ago to launch Xiaodu, a hardware startup that builds Amazon Echo-like smart devices.

“Eric and Kay are seasoned serial entrepreneurs with a proven track record of building successful products and companies, especially in the areas of AI and search,” Tan told JS. “Their team’s extensive experience uniquely positions them to drive breakthrough innovations.”

But I think it’s an uphill battle.

Assuming Genspark can work out the teething problems of its technology, identify a revenue-generating plan, and scale its small (~20-person) Singapore and Bay Area-based team, none of which are easy tasks, it will face intense competitive pressure from rival startups with hundreds of millions of dollars in the bank – not to mention established search companies like Google.

Genspark can do that too Real survive the bad optics and failed go-to-market strategies that have plagued other attempts at AI-powered search engines? And can it conquer a niche in a future where, for example, OpenAI launches a similar instrument?

I’m not convinced. But Jing is convinced it can be done.

“Many internet users, especially those younger than Google, don’t want to just be given a list of links and have to figure out the rest on their own while navigating sponsored content and SEO-driven content that games the system. Jing said. “They want to find what they need faster, they want more visual results and they want to know that the results are reliable. With AI we can achieve all that, and we launched Genspark to meet those needs.”