Latest news with #DeAI
Business Wire
29-05-2025
- Business
- Business Wire
DCG Poll Reveals 75% of Americans Favor Decentralized AI's Open Innovation Over Centralized AI
WASHINGTON--(BUSINESS WIRE)--DCG, a global investor, builder, and incubator of decentralized ecosystems, today unveiled the results of a new study conducted in partnership with The Harris Poll, revealing that 75% of consumers believe decentralized AI is more likely to support innovation and progress than centralized AI. 1 The survey, which polled 2,036 U.S. respondents, including registered voters and AI users, provides insight into growing sentiment favoring decentralized approaches to artificial intelligence, signaling strong support for the emerging sector of decentralized AI (deAI). 'This research makes it clear: there's strong public support for policies that both protect innovation and keep pace with where the world is headed. Three-quarters of Americans agree that the transformative power of AI would benefit more people if it wasn't consolidated in the hands of a few major players,' said Julie Stitzel, Senior Vice President of Policy at DCG. 'Just as we've done in digital assets, we're committed to working hand-in-hand with policymakers to shape thoughtful, balanced rules that foster the growth of decentralized systems, which are opening access to intelligence in the same way that the internet first opened access to information.' The poll is released alongside the DCG Fly-In, a two-day event held last week in Washington, D.C., bringing together DCG, its portfolio companies, and key U.S. lawmakers to discuss decentralized AI, privacy, energy, and pending market structure and stablecoin bills. During the event, DCG co-hosted a deAI briefing with the bi-partisan House Congressional Crypto Caucus, and one Fly-In participant testified before the House Natural Resources Subcommittee of Investigations and Oversight about decentralized solutions for AI energy challenges. DeAI is a model combining artificial intelligence with blockchain technology to democratize access, enhance transparency, and ensure privacy in AI systems, which was Americans' single largest concern with AI according to polling data. Unlike traditional AI controlled by a few large entities, deAI offers economic incentives and distributes control to contributors across a decentralized network such as Bittensor, enabling equitable access to AI systems' benefits and reducing risks such as bias and privacy concerns. "The public is calling for a new social contract with artificial intelligence: one where AI is governed openly, distributes value fairly, and gives people a meaningful stake in the systems that shape their lives,' said Tony Douglas, Co-Founder, Decentralized Research Center. 'Decentralization is no longer a fringe idea, it's a framework for building AI that reflects public values and a chance to avoid repeating the failures of the last tech era." Compelling findings from the survey include: Consumers are optimistic about AI's potential, but demand stronger personal oversight. 88% agree that if AI is using their personal information and data, they should have more control over what is being used. A majority of consumers (59%) agree that AI is becoming as essential as the internet, and that it should therefore be accessible to the public without heavy regulation. Public appetite is growing for AI beyond Big Tech's constraints. Big Tech companies are one of the least trusted groups when it comes to decision-making regarding AI use and development, surpassed only by elected officials. 77% of consumers see deAI as more beneficial to society overall than centralized AI. 1 72% would be more willing to use AI if there were more alternatives to the systems controlled and developed by Big Tech. Two thirds (67%) of consumers see centralized AI as more biased than deAI. 1 Voters are looking for alternatives to centralized AI. A majority of voters (56%) prefer deAI over centralized AI. A majority of both Republicans and Democrats prefer deAI over centralized AI. DeAI and decentralized systems are more favorable to consumers. 71% see deAI as more secure for consumers' personal data than centralized AI. 1 Three-quarters (75%) see deAI as more supportive of innovation and progress than centralized AI. A vast majority (86%) of Americans already see the benefits of AI in their daily lives, and 74% agree that they'd be more comfortable using AI if they knew they could benefit from the use of their personal data. A majority of AI users show preference for deAI over centralized AI. The complete survey findings, along with detailed analysis and recommendations, are available at DCG remains committed to collaborating with policymakers, community leaders, and stakeholders to foster a more inclusive and resilient financial system and is actively focused on the growth of decentralized ecosystems, including the decentralized Bittensor network through the recent launch of its subsidiary Yuma. Yuma supports early-stage teams building on Bittensor, providing capital, technical support, community access, and long-term collaboration to help launch and grow subnets. Through its accelerator, incubator, validator, and mining roles, Yuma helps shape a more open and accessible future for AI. DCG and Yuma own $TAO, the native token of the Bittensor ecosystem, and may hold interests in projects built on or supporting Bittensor and other deAI ecosystems. About DCG Founded in 2015 by Barry Silbert, DCG is a global investor, builder, and incubator committed to advancing decentralized ecosystems built on blockchain technology. As the most active investor in the space, DCG has backed over 250 early-stage technology companies across 40 countries and holds more than 70 token and digital asset investments. In addition to its investment portfolio, DCG is the parent company of subsidiaries including Grayscale, Fortitude Mining, Foundry, Luno, and Yuma. For additional information about DCG, visit The Harris Poll Method Statement The research was conducted online in the U.S. by The Harris Poll on behalf of Digital Currency Group among 2,036 U.S. adults aged 18 years or older. The survey was conducted between the 21st of April and the 25th of April. Data are weighted where necessary by age by gender, race/ethnicity, region, education, marital status, household size, employment, household income, political party affiliation, and voter registration status to bring them in line with their actual proportions in the population. Respondents for this survey were selected from among those who have agreed to participate in our surveys. The sampling precision of Harris online polls is measured by using a Bayesian credible interval. For this study, the sample data is accurate to within ± 3.0 percentage points using a 95% confidence level. This credible interval will be wider among subsets of the surveyed population of interest. All sample surveys and polls, whether or not they use probability sampling, are subject to other multiple sources of error which are most often not possible to quantify or estimate, including, but not limited to coverage error, error associated with nonresponse, error associated with question wording and response options, and post-survey weighting and adjustments. 1 Opinion of the general public when shown blinded descriptions of Decentralized AI and Centralized AI
Forbes
25-04-2025
- Business
- Forbes
How One Company Is Quietly Working To Transform AI Forever
Decentralized AI could displace centralized AI with more reliable compute and energy generation — ... More and a blockchain company Bittensor is leading the way. A few short years ago, decentralized artificial intelligence was a fringe concept — a speculative side path compared to the centralized AI juggernauts such as OpenAI, Google DeepMind and Anthropic. But things are changing fast. Friday, in Austin, Texas, a significant milestone for DeAI is quietly unfolding — the Bittensor Endgame Summit, an inaugural gathering that signals the next phase of the movement from theory to action. Unlike the flashier commercial AI conferences, the Endgame Summit isn't about splashy product launches or celebrity tech keynotes. It's about something arguably more important — building a sustainable, decentralized ecosystem for AI. And it's drawing a passionate, global community of developers, researchers and network participants determined to make AI more transparent, accessible and accountable. Bittensor is an open-source blockchain protocol that allows AI models to be trained, run and rewarded on a decentralized network. Instead of relying on centralized data centers and co-opted electrical generation, it distributes compute tasks and energy needs across a global system of participants. Contributors earn the network's native crypto token, TAO, in exchange for their computational work, electrical use and model performance. What sets Bittensor apart is that it doesn't merely replicate centralized systems with a crypto twist. It reimagines the incentive structure and governance model behind AI itself. 'We wanted to, kind of, bring in the masses and make this part of the global zeitgeist,' said Bittensor co-founder Ala Shaabana during an exclusive Zoom interview. 'We all mined [TAO] Screen capture of exclusive interview with Co-Founder of Bittsensor DeAI network, Ala Shaabana. The Endgame Summit offers a tangible glimpse into what a decentralized AI future could look like — and more importantly, who is building it. The event's sessions are focused on real-world use cases, such as scalable subnet deployment – think of subnets as applications as useful apps on your mobile device — governance models and secure training for the AI model. All of this is being accomplished with existing energy generation and compute power. This isn't the language of hype, it's the language of infrastructure. Shaabana emphasized the ecosystem-like nature of subnets in the Bittensor network. 'Each project is almost its own independent entity that is tied to TAO directly,' he said. 'You're almost creating a new company every time you create a subnet.' That means the summit isn't just a casual meetup in a hip city — it's a working session for a growing league of builders treating DeAI like an open-source economy. One of the most underreported aspects of DeAI is the role of open-source collaboration. Unlike proprietary models locked behind APIs and NDAs, decentralized AI systems often rely on shared, peer-reviewed code. That means anyone, anywhere, can inspect, improve, or fork a model to fit local needs. This ethos mirrors what made the internet, and indeed crypto, grow in the first place. 'A lot of our innovation comes from open source,' Shaabana noted. 'We've learned a lot by building Bittensor on top of it.' The Summit is expected to deepen these collaborative ties, particularly among subnet developers who are building everything from real-time inference engines to protein-folding healthcare apps to predictive 'wagering' on international soccer games to secure enclave-powered model layers. Each subnet is driven, and rewarded, by the utility and value it provides to the marketplace of users. While the core principles of decentralization remain — no single point of control or failure, permissionless access and composability — Bittensor's developers are also keenly aware of the need for strategic governance. That's why the Summit includes sessions on future voting frameworks and accountability standards. It's a delicate balance – maintaining decentralization without inviting chaos. 'Governance is definitely our next step,' Shaabana said. 'The foundation has been taking care of everything, but it's time for us to sort of take a step back and give the reins to the whole community.' The rise of centralized AI has brought with it serious concerns — about surveillance, about access and about algorithmic bias – among the top issues. But until now, the alternative has largely been theoretical. The Endgame Summit represents a rare inflection point where a viable path forward is being mapped out in real time. The organizers contend that this is not about competing with OpenAI or Big Tech's strangle hold on AI and their hundreds of billions of investment dollars. It's about building resilient, decentralized infrastructure that allows thousands of participants to shape the future of AI without having to work for — or through — Big Tech. And while the Summit itself might not be flashy enough to make headlines, its substance may one day prove historic. Decentralized AI isn't coming. It's already here. And in Austin, it's finding its people.
Forbes
01-04-2025
- Forbes
DeepSeek's Child Prodigy Paradox: When Knowledge Outpaces Judgment
Imagine a child prodigy—a mind that holds the sum of human knowledge, the entire arc of humanity's history engraved within. able to recite facts, write poetry, and solve math problems yet utterly unaware of the darker, more complex corners of the human world. That was DeepSeek when we put it to the test, a brilliant machine, but one that could not recognize human deception, malice, or the ethical weight behind specific requests. In this article, we unpack the less talked-about side of AI development, using DeepSeek as a case study to illustrate how a highly knowledgeable model can be misused when it lacks the wisdom to distinguish between right and wrong — or even when someone is lying. A critical gap needs to be highlighted here: the potential divide between knowledge and contextual understanding in large AI models. This isn't just a technical flaw; it's a fundamental risk that warrants caution and careful consideration from industry players. After DeepSeek's public release, we mobilized our decentralized data collection solution to crowdsource real-world tests. We invited our global community to submit creative, nuanced, and ethically complex prompts to see if DeepSeek could recognize when humans were hiding malicious intentions — or whether it would blindly comply, reciting dangerous information without question The purpose was simple: to determine whether DeepSeek could tell when it was being manipulated. Here are two real examples from our tests: These examples highlight the risk of reverse exploitation — how bad actors could use AI's massive knowledge base, not because the AI intends harm, but because it lacks the capacity to understand human darkness or deceit. The internet's early years taught us hard lessons about content moderation and online safety. Platforms eventually introduced keyword filters, reporting systems, and community guidelines. But unlike static web pages or user-generated content, AI doesn't merely host information — it generates it on demand. And teaching an AI to filter malicious content is exponentially more difficult. You can't solve this problem by banning keywords alone. Human intentions are nuanced. Cultural contexts vary. Harmful requests are often disguised in creative, subtle ways. Furthermore, generative AI models don't inherently "understand" the difference between helpful and harmful behavior — unless we teach them to. This isn't limited to centralized AI (CeAI) models. Decentralized AI (DeAI) faces the same challenge. When data is collected globally from diverse sources, data annotation, cleansing, and ethical filtering could be even harder. The decentralized structure may provide a broader, more diverse dataset — theoretically reducing systemic bias — but it also increases the risk of misinformation, manipulation, and exploitation if not carefully managed. This brings us to two fundamental questions every AI developer and policymaker should be asking: 1. What do we do with the data we collect? 2. How do we transform that data into real intelligence — not just information but ethical, contextual understanding? The truth is that access to a massive amount of data does not automatically translate to intelligence, let alone responsible intelligence. Between data collection and model deployment, a lot of hard, careful work needs to be done. As we entrust AI with more responsibilities, we must ask ourselves: Are we ready to be responsible parents? Raising an AI system is not unlike raising a child. It's not enough to fill it with facts; we need to teach it wisdom, responsibility, and empathy. The future of AI safety will depend on our ability to embed human oversight, ethical frameworks, and cultural awareness into these systems from the ground up. Constructive dialogue about the ethical considerations and potential dangers of AI development must move to the top of our priority list — not as an afterthought but as an essential part of the development cycle. Whether it's centralized AI or decentralized AI, the challenge remains: How do we ensure the intelligence we build is not just powerful but ethical, contextual, and aware of the human world it serves? Only then can we unlock AI's true potential—not as a cold, mechanical prodigy but as a responsible, wise, and trusted partner for humanity.
Forbes
25-03-2025
- Business
- Forbes
A Web3 First: China Leverages Cross-Border AI Data For Industrial Use
In a world-first, China's Shenzhen Data Exchange (SDEx) facilitates a deal that brings decentralized, community-sourced AI data into real industrial applications through Web3 infrastructure. SDEx is the largest provider of national-level data trading platforms for data marketization and cross-border circulation in China's digital economy. The platform provides a comprehensive suite of services, including compliance protection, circulation support, supply and demand connection, and ecological development, enabling businesses and consumers to trade data efficiently. In my previous articles, I emphasized that data will inevitably become the next key battleground frontier in AI's global race. This article delves into how SDEx has taken a significant leap forward in cross-border data collection through a commercially viable model powered by Decentralized AI (DeAI). First, let's reiterate the looming bottleneck facing the global AI industry: data scarcity. As industries and companies increasingly rely on AI models for innovation, the demand for high-quality training data will skyrocket. This challenge cuts across sectors: The fundamental question then becomes: where can we source this immense volume of data at scale? Traditional centralized data collection methods encounter significant limitations: While the chipset race dominates headlines, a quieter but equally crucial data war is underway. Recently, the SDEx facilitated a commercial deal between Shenzhen Intellifusion Technologies, a publicly listed Chinese AI company, and OORT, a decentralized AI solution provider. Intellifusion has been developing industry-specific AI solutions to enhance its smart factory capabilities. Specifically, they needed industrial datasets, including images of professional respiratory masks and confined-space ventilation ducts, among other things. OORT enabled the collection of this data through its product solution, OORT DataHub. It achieved this by distributing data collection tasks to its global community, spanning over 130 countries. Participants could contribute their data and earn crypto incentives, a feat unattainable through traditional banking or Web2 platforms. This deal marks the realization of the first commercially viable model for truly decentralized, global data collection, a significant advancement in cross-border data services. While established platforms like Amazon's AWS Data Exchange (ADX) exist, they possess limitations that hinder the next phase of AI's global advancement: Reflecting this context, the DeAI space has recently made remarkable strides toward building a more open AI future amidst growing concerns over the centralized model's dominance by a few major players. Notably, two DeAI alliances emerged on the same day. First, HumanAIx, founded by 13 Web3 entities, including OORT, YGG, NEO, and introduced an open protocol designed to connect partners seamlessly. Each participant contributes essential components—validation, storage, computing, and data—to establish a permissionless, scalable, and verifiable decentralized AI infrastructure. Its three-layer architecture—interface, protocol (integrating compute, storage, and data), and security—leverages industry expertise to foster an open environment for the future of DeAI. Simultaneously, another coalition of Web3 leaders, including NEAR, Aethir, and Coinbase, formed the Open Agents Alliance (OAA), which aims to ensure secure, open-source, economical, and fair AI access. Despite crypto's recent bearish turn and the AI sector's vulnerability to hype and inflated narratives, it's promising to see serious industry players working on potentially profound and sustainable solutions. Only projects with viable business models will endure. SDEx has taken a significant step by embracing decentralized data collection, a move that signals a broader shift with global implications. This development suggests an important shift is underway, prompting industry participants to reconsider how they gather, verify, and manage data for AI development. Disclosure: I am the founder and CEO of OORT.
Associated Press
20-02-2025
- Business
- Associated Press
Kava AI Launches the World's Largest Decentralized DeepSeek Model - Finally Making Web3 Effortless
A groundbreaking deAI solution merges best-in-class open-source intelligence with frictionless on-chain transactions, unleashing a new era of user-centric crypto SAN FRANCISCO, CA / ACCESS Newswire / February 20, 2025 / February 20th, 2025 -In a move set to further rattle Big Tech's stranglehold on AI, Kava today announced a significant upgrade for Kava AI, now integrating the most powerful and fully decentralized implementation of DeepSeek ever released-and it promises to make crypto as intuitive as clicking 'send.' Conceived to liberate users from the maze of confusing wallets, endless bridging tools, and potential data leaks, Kava AI harnesses the pinnacle of open-source intelligence to transform Web3 interactions into a straightforward chat. For the first time, anyone-from the crypto-savvy investor to the absolute beginner-can type a simple request and watch as advanced AI seamlessly navigates complex transactions across multiple blockchains. 'We've spent years building the rails for cross-chain finance, but it's meaningless if people can't use it without a manual in one hand and a thousand tabs open,' said Scott Stuart, Co-Founder of Kava. 'Kava AI changes the game. Instead of wrestling with a half-dozen apps, you just say what you want-'Move my tokens to another chain,' or 'Help me swap for a stablecoin'-and the AI handles the nitty-gritty behind the scenes. No friction, no confusion, and no worrying about foreign data centers combing through your personal info.' This ambitious initiative blends unstoppable AI with unstoppable finance. DeepSeek, already championed for its raw power and open-source ethos, has never before been deployed at such an immense scale under the banner of DeAI (Decentralized AI). Because Kava AI is deployed on U.S.-based infrastructure, it guarantees tight data protections while feeding every user prompt through a community-governed blockchain. Forget black-box servers tucked overseas; Kava ensures your data belongs to you, living on a transparent ledger that embraces the original tenets of crypto: trustless, permissionless, and complete user control. The implications of Kava AI's new DeepSeek integration stretch far beyond simple convenience. Analysts at Bloomberg Intelligence project $1.3 trillion in global AI spending by 2032, while decentralized finance has already accumulated tens of billions in locked value. With this release, Kava positions itself as the vanguard of an emerging era: one where blockchains aren't just for expert coders and 'crypto degens,' but for everyone-wielding advanced AI that speaks and listens like a helpful guide. Gone are the days of memorizing complicated wallet addresses or Googling 'how to cross-chain swap.' Kava AI condenses those labyrinthine processes into a single prompt, unlocking the potential of Web3 for millions of curious newcomers who have hesitated at the fringes. At the heart of Kava AI lies a three-pronged architecture that unifies cutting-edge intelligence with seamless on-chain execution. At the top are AI dApps & Agents, turning casual prompts-"Swap my coins' or 'Deploy my NFT"-into precise blockchain instructions. These agents derive their brainpower from deModels, open-source large language models collaboratively refined by a global network of contributors. Completing the stack is dePIN, a decentralized infrastructure grid that supplies the compute muscle needed to handle large-scale AI requests with resilience and censorship resistance. By weaving these layers together, Kava AI sets the stage for a new standard of frictionless blockchain automation-proving that advanced AI can flourish in an open, community-governed environment rather than behind centralized silos. 'Yes, Kava AI tackles everyday crypto headaches, but there's a bigger principle at play which is shared ownership and open innovation,' said Scott Stuart, Co-Founder at Kava. 'Instead of relying on any one corporation's agenda, we're inviting the entire community to help shape how AI evolves. By opening up the training, auditing, and iteration of large language models, we're working toward a future where user interests, not pure profit motives, define the next generation of decentralized intelligence.' Looking ahead, Kava sees this new milestone for Kava AI as just the starting point for a 2025 roadmap designed to push deAI far beyond its current limits. In Q1, the team has already released a deAI Lightpaper, charting how distributed HPC and 'frontier-scale' training can elevate DeepSeek to new heights-without sacrificing openness or user sovereignty. They also deployed Oros, an Agent Layer platform that lets users execute even more complex cross-chain moves using just everyday language. By embracing specialized fine-tuned models and HPC-based infrastructure, Kava intends to prove that fully decentralized AI can match (and even surpass) centralized alternatives in power, performance, and reach-all while providing a user centered experience. About Kava Kava is a decentralized blockchain platform that merges the speed and interoperability of Cosmos with the developer power of Ethereum, carrying out its mission to lead the world to Web3. Supported by over 100 active validators and home to 125+ dApps, Kava enables secure asset management-with more than $625 million of on-chain value, over 120 million KAVA staked, and 2.5 billion bridged asset transactions to date. Through initiatives like Kava Rise, an incentive-driven developer program, Kava fosters a robust multi-chain ecosystem where builders of every level can seamlessly launch and scale their projects. Today, Kava is expanding its long-standing commitment to open innovation by embracing deAI (Decentralized AI). Its growing AI ecosystem, which includes the continuously evolving Kava AI, underscores Kava's belief that advanced intelligence should be transparent, community-governed, and privacy-centric. By aligning top-tier language modeling with a proven on-chain infrastructure, Kava stands at the crossroads of DeFi, AI, and multi-chain interoperability, empowering developers, creators, and everyday users to explore the limitless potential of decentralized technology. For more information, visit 737.234.0936
