Dozens of builders and technology enthusiasts gathered at DKGCon 2024 from Oct. 24–25 to explore the intersection of blockchain, artificial intelligence and knowledge graphs, with a particular focus on how Web3 tools can improve the efficiency, safety and reliability of AI.
The event was organized by OriginTrail, a blockchain ecosystem focused on building what it describes as the world’s first decentralized knowledge graph, or DKG, which it says helps to bring trust and reliability to AI.
The agenda included talks and panels related to user privacy and empowerment as well as the impact of AI on various industries, with speakers ranging from Ethernet inventor Bob Metcalfe and prominent AI researcher and critic Gary Marcus to SingularityNET CEO Ben Goertzel.
There was a particular focus on ecosystem builders, with a dedicated section on day one for people to present the projects they are building using the DKG and a second conference day entirely for builder workshops.
Attendees learn how to build using the DKG on day two of the conference. Source: Jonathan DeYoung
The OriginTrail team told Cointelegraph that the conference is “an annual celebration of advancements in the field of decentralized knowledge graphs, powered by the OriginTrail ecosystem. This year, the focus was on incentivizing collective neuro-symbolic AI, showing the strength of synergy between knowledge graphs, blockchains and artificial intelligence.”
“Trust the source”Knowledge graphs are a type of data layer that helps establish and verify relationships between various data points. They are frequently used by Big Tech companies like Google and Facebook to power their recommendation algorithms, but they typically remain closed ecosystems, where centralized companies control the data and users do not share in its monetization.
Knowledge graphs are also becoming increasingly popular within large language models (LLM) and generative AI applications because they help to solve issues such as hallucinations and intellectual property infringement. The OriginTrail team told Cointelegraph that by decentralizing knowledge graph technology, users are even further empowered:
“Instead of hallucinations, we can have information provenance; instead of bias, we can have reliability; and instead of IP infringements, we can incentivize data ownership.”As such, the conference’s theme was “trust the source,” a slogan printed on the back of T-shirts handed out to attendees in their swag bags. According to the team, the phrase references the “critical challenge of achieving transparency of AI solutions.”
OriginTrail says the DKG helps you “trust the source” in AI applications. Source: Ema Lovšin
“With the ambition of AI impacting almost every aspect of our modern lives, we need to make it a reliable, trusted and transparent system,” said the OriginTrail team. “The tagline highlights the capability of the Decentralized Knowledge Graph to enable users to trust the source of information when using AI.”
Ethernet inventor Metcalfe opened up the conference and told the audience that “LLMs and knowledge graphs are made for each other. They solve each other’s pathologies.” The pathologies he was referring to were AI’s tendency to hallucinate and spit out incorrect or biased information.
Related: OriginTrail on AI, real-world adoption and the value of knowledge
Several non-Web3 organizations also shared the ways the combination of blockchain with the DKG has improved trust for users. During a panel titled “Is AI ready for business? Breaking down trust and reliability barriers,” Garrett Pitcher, founder of Irish whisky distillery Church of Oak, shared that it allows customers to reliably trace the origins of the products they purchase.
Meanwhile, Seamus Galvin, senior innovation business partner at international standards institute GS1 — the organization behind the barcode — said that “it’s the convergence of blockchain, with AI, with semantic web” that achieves “supply chain trust [and] supply chain integrity.”
AI safety and privacyAnother theme permeating the conference was user safety and privacy in AI applications. Several speakers spoke on the idea of the need to build guardrails into AI systems so that they do not harm or exploit users. This concern is underscored by the recent death of a teenager who reportedly took his own life after an AI chatbot encouraged him to do so.
There are also concerns about what centralized AI companies are doing with the vast amounts of user data they consume. OpenAI, in particular, has come under scrutiny and was famously grilled by AI critic Marcus in 2023. During a fireside chat at DKGCon 2024, Marcus argued that due to its precarious financial situation, OpenAI would have no choice but to become a surveillance company:
“I think OpenAI is going to be forced to become a surveillance company. I think it’s going to become the most Orwellian name ever. […] I think the only thing they can do is surveillance, and I think there are a lot of signs they are moving towards that.”Gary Marcus (right) in a virtual fireside chat with OriginTrail co-founder Branimir Rakić (left). Source: OriginTrail
This sentiment was also shared by American whistleblower and privacy advocate Edward Snowden after the firm appointed former National Security Agency Director Paul Nakasone to its board in June 2024. Snowden wrote on X that OpenAi had gone “full mask-off, adding, “Do not ever trust @OpenAI or its products (ChatGPT etc).”
Likewise, AMYP Ventures managing partner Chris Rynning — who helps manage the wealth of the Porsche family — said during a panel titled “Catalyzing the future: Impact of AI on the startup ecosystem” that he had been reluctant to invest in AI startups due to concerns around “funding the potential end of humanity.”
Building AI tools that won’t exploit usVarious speakers at the conference positioned the DKG as a safe, private solution to these concerns and an alternative to what many see as Big Tech’s data overreach and reckless approach to developing AI. The OriginTrail team told Cointelegraph that combining AI with Web3 and knowledge graphs offers “humanity’s best bet to reach AI’s full potential safely.”
OriginTrail co-founder and chief technology officer Branimir Rakić told the audience that “we’re taking it in a completely opposite direction,” pointing to its Edge Node application, which allows users to interact with AI services while keeping their data private on their local device.
Later, Rakić joined OriginTrail’s other founders to close out the main event by arguing the mainstream AI industry needs a decentralized alternative, as “surveillance capitalism” is the status quo. “It’s really a chase for who can get more data and more chips to like squeeze that data in and train a model that ultimately is questionable of not just how well it performs, like what it ends up doing.”
He went on to describe OriginTrail’s philosophy as “a counter to that,” adding:
“We really need to go back to this power-to-the-people approach if we’re ever going to take on this hugely powerful technology.”The attendees Cointelegraph spoke with at the event all seemed to agree that decentralizing the control over AI through the convergence of knowledge graphs and blockchain was the best and only way forward to offer a meaningful counterbalance to Big Tech’s approach. And when the stakes are as high as the potential annihilation of humanity, the era of AI may be Web3’s greatest — and most important — opportunity to disrupt the status quo.
AI Eye: A bizarre cult is growing around AI-created memecoin ‘religions’
Disclaimer: OriginTrail paid Jonathan DeYoung’s travel expenses to attend DKGcon 2024.
{Categories} _Category: Implications{/Categories}
{URL}https://cointelegraph.com/news/decentralized-ai-trust-privacy-dkgcon-2024?utm_source=rss_feed&utm_medium=rss&utm_campaign=rss_partner_inbound{/URL}
{Author}Cointelegraph by Jonathan DeYoung{/Author}
{Image}https://images.cointelegraph.com/images/840_aHR0cHM6Ly9zMy5jb2ludGVsZWdyYXBoLmNvbS91cGxvYWRzLzIwMjQtMTAvMDE5MmQ5NWUtZWNjMS03NDY4LTlmMDktYmUyNmQxM2QzNDJm.jpg{/Image}
{Keywords}OriginTrail,DKGcon,DKG,decentralized knowledge graph knowledge graphs,AI,artificial intelligence,privacy,OpenAI{/Keywords}
{Source}Implications{/Source}
{Thumb}{/Thumb}