Venice AI, the distributed generative AI model from ShapeShift founder Erik Voorhees, took a big leap, announcing on Thursday that its service is now web-connected and capable of real-time responses.
Voorhees launched the privacy-focused Venice AI in May as an alternative to centralized, censored AI models like OpenAI's ChatGPT, Anthropic's Claude AI and Google's Gemini, billing the project as “a team of crypto people building AI apps designed in the spirit of crypto.”
“As always, and unlike ChatGPT/Claude/Gemini, all conversations on Venice are private and permissionless,” Voorhees wrote on Twitter about the new search feature. “Venice cannot see your messages or responses, and all data is stored locally in your browser. Venice does not censor AI answers.”
Voorhees said he is concerned that AI will become too centralized and controlled by tech giants, necessitating the need for decentralized services.
“I saw where AI was going, and it was going to be taken over by big tech companies in collusion with governments,” Voorhees previously said. Decryption“And that really worried me. And I realized how powerful AI is and how profound an impact it can have. This is an amazing area of emerging technology.”
But before the update, Venice AI's responses were limited to knowledge from two years ago, lagging behind last year's ChatGPT and Gemini search updates, which provided real-time results.
“We exclusively use the open source model, and the open source model is always limited by publication dates,” said Tiana Baker Taylor, chief operating officer at Venice AI. Decryption“The two models we're looking at right now are both versions of Meta's Llama-3, and they're great models, but they were released in 2022. So we knew it was going to be a challenge for information querying the models needing data from 2022 to the present.”
As Baker Taylor explained, a key feature that accompanies the search is Venice AI's ability to cite sources and provide links to the information provided.
“I think that's really important when people are questioning something, checking facts, or trying to learn new information. [that they] “You can't tell where the information comes from,” she says. “The model itself is open source, so you can see what data it was trained on and what weights were given to different parameters.”
Faster, more accurate responses are important for any generative AI model, but Baker Taylor says openness is key.
“This increases transparency because people know exactly where the information is coming from and it is cited, which increases confidence in the accuracy of the information,” she said.
Baker-Taylor explained that Venice AI uses Brave's API for its searches, and said that as a privacy-focused browser company, Brave was a natural choice. He noted how Venice AI is staying true to its core vision:
“There are a few fundamental principles on which we're building Venice,” Baker Taylor said. “The first is privacy. Privacy is a human right, and people should have access to machine learning and artificial intelligence, but they should also have the ability to access it without the conversations and content they create being tied to their identity and potentially searchable, reported and sold for the rest of their lives.”
Baker-Taylor also emphasized his goal of keeping Venice AI uncensored.
“Another key foundational principle is that access to machine learning should not be censored, and when you interact with these technologies you should receive as close to the raw machine learning intelligence as possible, without the added human bias that you might find on other platforms.”
Baker-Taylor added that in addition to the search functionality, new features being introduced on Thursday include a share button that allows users to share prompts or messages with others or themselves using an encrypted link.
“Our privacy infrastructure is device-centric, so everything is stored in your browser, not on our servers,” Baker Taylor says, “which means if you're having one conversation on your phone and another on your desktop, the twain don't match up.”
“This will allow you to share chat conversations,” she added, noting that by clicking the link, the receiving user will be able to resume the conversation from where the previous user left off.
“We think people are going to use this in a lot of different ways because they don't have to reinvent the wheel. We're really excited about it,” she concluded.
Editor: Ryan Ozawa.
