AI candidate running for Parliament in the U.K. says AI can humanize politics
AI News Voters can interact with AI Steve, a candidate in next month’s U.K. general election, to discuss policy issues or voice their concerns.
In the upcoming general election in the United Kingdom, an artificial intelligence named “AI Steve” will be listed on the ballot in the Brighton Pavilion constituency of Brighton and Hove. Represented by businessman Steve Endacott, AI Steve is not just a campaign tool but a symbol of integrating AI into politics. Endacott, the chairman of Neural Voice—a company specializing in AI voice assistants—describes AI Steve as his “AI co-pilot,” indicating that while he is the official candidate, his campaign decisions are influenced by AI.
AI Steve is part of a broader initiative by Neural Voice to demonstrate the potential of AI in daily interactions and governance. Constituents can interact with AI Steve via its website, asking questions about policy or offering opinions, to which AI Steve responds with information drawn from a detailed database of the party’s policies. If the AI lacks a predefined policy response, it conducts online research to provide an informed answer and even engages users to contribute their policy ideas.
This innovative approach allows AI Steve to provide constant availability to constituents, promising a new level of responsiveness in political representation. In a recent interaction, AI Steve discussed the UK’s decision on Brexit, stating its role to implement and optimize the decision, reflecting its programming to abide by democratic outcomes while inviting public input on managing future implications of Brexit. This example underscores the experimental nature of AI Steve’s candidacy, blending technology and traditional political engagement to potentially reshape public interaction in government processes. AI News
Steve Endacott, leveraging his candidacy in Brighton Pavilion, is seeking a unique approach to policymaking by engaging “validators,” whom he considers representative of the average constituent. These validators, particularly locals with lengthy daily commutes, are asked to score proposed policies on a scale from 1 to 10 on a weekly basis. Any policy receiving a score above 50% is then adopted as the official party stance. Endacott emphasizes that his policy decisions are directly influenced by his constituents’ preferences, which he gathers electronically each week, ensuring continual voter engagement.
AI News In a previous 2022 local election as a Conservative Party candidate, Endacott garnered less than 500 votes. However, his latest campaign, marked by the introduction of the AI-driven initiative “AI Steve,” has sparked significant interest. This was particularly evident over one weekend when details about AI Steve were inadvertently released online, leading to approximately 1,000 interactions with the AI platform in a single night, highlighting the public’s curiosity and engagement with this innovative campaign strategy.
Steve Endacott, campaigning in the Brighton Pavilion area, is utilizing an AI tool named “AI Steve” to field concerns from potential constituents. This innovative approach has allowed him to efficiently gather voter priorities, which so far include Palestinian safety, local sanitation, bicycle lanes, immigration, and abortion rights. By leveraging AI, Endacott can interact with thousands of voters daily, a method he describes as a modern take on political engagement, allowing constituents to initiate contact at their convenience rather than through traditional door-to-door canvassing.
Endacott identifies as a “centralist,” closely aligning with but not fully endorsing the Green Party. His new party, Smarter U.K., didn’t register in time for this election, but he remains committed to advocating for significant reductions in carbon emissions, aiming to influence policy either through election or as a political influencer.
His involvement with Neural River, the company behind AI Steve, is minimal, as he holds less than a 10% share, underscoring that his campaign is not driven by personal business interests. This initiative aligns with global trends where candidates are exploring AI’s role in decision-making processes, as seen in similar moves by political figures in Wyoming, USA, and Denmark.
Endacott emphasizes that his use of AI in politics is a serious endeavor aimed at increasing public involvement in democratic processes, not just during election cycles but on a continuous basis. This approach, he argues, is a transformative way to enhance democratic engagement, making politics more accessible and responsive to the electorate’s needs. SOURCE AI News
Exclusive: Former Meta engineers launch Jace, an AI agent that works independently
AI News Zeta Labs, a startup based in London and established by ex-Meta engineers Fryderyk Wiatrowski and Peter Albert, has introduced Jace, an AI agent powered by large language models (LLM). Jace is capable of performing tasks directly within a web browser based on user commands. Alongside the launch, Zeta Labs announced a successful pre-seed funding round, raising $2.9 million. The round was spearheaded by notable investors including Daniel Gross, former head of AI at Y Combinator, and Nat Friedman, former CEO of GitHub. This funding will support the further development and deployment of Jace, enhancing its in-browser functionality and expanding its potential applications.
Amidst the growing interest in AI agents like Cognition’s popular Devin, Zeta Labs asserts that its AI agent, Jace, distinguishes itself by operating without user guidance, offering users the convenience of managing tasks without being physically present at their computers. Users simply instruct Jace on the desired tasks, and the agent autonomously executes them. Currently, Zeta Labs is collaborating with select early partners to refine Jace’s functionality, with the aim of enhancing its speed and reliability for handling complex tasks required by both consumers and businesses.
The startup’s recent pre-seed funding round, amounting to $2.9 million, is set to further this development. The investment was led by notable figures such as Daniel Gross and Nat Friedman, and supported by a group of angel investors and venture capital firms, including Shawn Wang, Bartek Pucek, and Mati Staniszewski, the founder of ElevenLabs. This financial backing will support Zeta Labs in advancing Jace’s capabilities to meet the increasing demands of its users.
Albert, inspired by his previous experience running an ecommerce business where mundane operational tasks such as data transfer consumed substantial time, recognized the potential for an AI agent to streamline such processes. This insight came over eight years ago, well before the advent of advanced language models. As AI technology evolved, particularly with the development of sophisticated GPT language models, Albert saw an opportunity to actualize his vision. He partnered with Wiatrowski, a fellow engineer from Meta, to found Zeta Labs. Together, they developed Jace, their core product designed to automate repetitive tasks that typically require human intervention in business operations. Jace leverages the capabilities of modern language models to perform a wide range of tasks, from data manipulation and integration across various platforms to more complex administrative functions, effectively reducing the workload and increasing efficiency in business environments.
Jace operates as an advanced web-based AI agent, similar in interaction to chatbots like ChatGPT, but with significantly enhanced capabilities. Users initiate tasks by entering instructions into a chat interface, where they can communicate naturally or respond to prompts that guide the task structuring process. Once Jace understands the task, it leverages sophisticated algorithms to plan and execute actions directly within the browser.
For example, if a user requests a hotel booking in Paris for a specific week, Jace does not merely gather information like a standard search agent; it actively accesses the hotel’s booking site, navigates the interface, inputs details, and completes the reservation, including handling payment transactions. This level of integration allows Jace to perform a wide array of web-based tasks—ranging from simple daily chores like searching for flight options or managing email correspondences to more complex activities such as setting up recruitment processes on LinkedIn, managing business inventories, or initiating marketing campaigns.
Albert, co-founder of Zeta Labs, explains that Jace essentially extends the functionality of text-generating AI by providing it with the capability to interact with and manipulate web environments. This makes it an extremely versatile tool that can automate operations across different platforms, significantly reducing the manual effort required for both mundane and complex tasks.
One of Jace’s notable achievements, as highlighted by Albert in an interview with VentureBeat, was its ability to autonomously establish a company. The AI navigated through the processes of drafting a business plan, registering the company, and securing its first client, thereby generating revenue. This example underscores the potential of Jace to perform autonomously in a commercial setting, showcasing how AI can extend beyond mere assistance to actively driving business initiatives and operations.
As Jace executes tasks, users have the option to switch the interface layout to observe the AI’s actions directly within the browser. This feature provides transparency and allows users to monitor Jace’s real-time web interactions, ensuring that every step of the task process is visible and trackable. This visibility not only enhances user trust but also allows for immediate feedback and adjustments if necessary, making the AI’s operation more interactive and comprehensible.
Jace, the autonomous web agent developed by Zeta Labs, incorporates a sophisticated blend of technological models to interact with and navigate the web. The system starts with a leading large language model (LLM) responsible for engaging in dialogue with users, gathering necessary information, and developing an actionable plan. This chat-based interaction model captures user intents effectively, translating complex requests into structured tasks.
AI News To translate these plans into direct web actions, Jace utilizes Zeta Labs’ proprietary model, known as AWA-1 (Autonomous Web Agent-1). This specialized web-interaction model takes the structured plans and executes them within web browsers, adeptly managing the inconsistencies and challenges posed by varying web interfaces. AWA-1 ensures that Jace can perform multi-step web tasks seamlessly, from booking hotels to setting up business processes online.
Fryderyk Wiatrowski, co-founder of Zeta Labs, explained that their core AI model is an enhanced version of an open-source LLM, augmented through a process called reinforcement learning from AI feedback (RLAIF). This approach involves intensive training with simulated interactions and synthetic data, preparing Jace to handle complex, multi-step tasks reliably.
To address potential issues where web agents may loop or stall, especially in tasks involving ten or more steps, Jace incorporates advanced reasoning systems. These systems check and confirm the completion of each step in the task plan, allowing the agent to handle tasks with potentially hundreds of steps without losing track.
Security is also a critical feature of Jace’s design. The agent employs stringent security measures to protect user data, such as encrypting credentials needed for tasks (like posting a job on LinkedIn) in a manner akin to secure password stores. This ensures that sensitive information remains protected while enabling Jace to perform its designated tasks effectively.
Zeta Labs is actively developing its AI agent, Jace, focusing on enhancing its capabilities and preparing for a broader market release. While Jace is already capable of performing a variety of web-based tasks, the company is not yet monetizing the agent. Current efforts are concentrated on refining the technology through collaborations with design partners, and significant advancements are being made toward the next version of the Autonomous Web Agent (AWA) model. This upcoming iteration promises to be larger, faster, and more adept at managing complex tasks that involve visual interactions, such as navigating maps.
The bulk of Zeta Labs’ pre-seed funding is allocated towards these developments, including the expansion of the team to support these efforts. Looking forward, Zeta Labs envisions Jace as a valuable tool for consumers and small businesses alike, automating mundane, browser-based tasks across various sectors like recruitment, ecommerce, marketing, and sales.
To make Jace accessible, Zeta Labs plans to introduce a tiered pricing model. Initially, users will have access to a free plan that includes a limited number of interactions. Once this cap is reached, continued usage will require a subscription, priced at $45 per month. This model is designed to make advanced AI assistance affordable while scaling up to meet the needs of professional users.
Fryderyk Wiatrowski highlighted the potential impact of Jace on small businesses, particularly in the recruitment sector. He described the current manual processes that recruiters undergo—such as using binary search strings on LinkedIn, transferring data into Airtable, and conducting matching analyses—which are time-consuming and error-prone. Jace aims to streamline these tasks by automating the entire pipeline, allowing recruiters to simply specify their needs and let the AI handle the rest, enhancing efficiency and accuracy in their workflows. SOURCE AI News
OpenAI appoints Retired U.S. Army General Paul M. Nakasone to Board of Directors
AI News Paul M. Nakasone, with his extensive background in cybersecurity, is set to enhance the Board of Directors at OpenAI, focusing specifically on issues of safety and security. His appointment includes a key role on the Board’s Safety and Security Committee, where his expertise is expected to significantly contribute to the organization’s strategic initiatives in these critical areas. This move aligns with OpenAI’s commitment to reinforcing its governance framework in response to the increasingly complex cybersecurity challenges facing AI development and deployment.
OpenAI is proud to announce the addition of Retired U.S. Army General Paul M. Nakasone to its Board of Directors. Recognized globally for his deep expertise in cybersecurity, General Nakasone’s appointment underscores OpenAI’s commitment to prioritizing safety and security amid the evolving challenges presented by AI technologies.
AI News General Nakasone will immediately take up a pivotal role on the Board’s Safety and Security Committee. This group is tasked with guiding the board on key safety and security strategies affecting OpenAI’s projects and operational frameworks. His involvement is expected to bolster the organization’s defenses against the complex cyber threats that accompany advancements in AI.
Central to OpenAI’s mission is the security of our operational infrastructure, from the supercomputers that power our AI training processes to the protection of sensitive data and model weights. As we advance toward the development of artificial general intelligence (AGI), enhancing our cybersecurity measures is crucial to withstand sophisticated threats.
Furthermore, General Nakasone will play a significant role in enhancing OpenAI’s exploration of AI’s role in cybersecurity. His expertise will aid in developing systems capable of rapidly detecting and neutralizing threats, potentially benefiting high-risk sectors such as healthcare, education, and finance.
Bret Taylor, Chair of OpenAI’s Board, emphasized the importance of robust security frameworks: “AI’s potential to positively impact society is immense, but this can only be realized through secure development and deployment. General Nakasone’s unparalleled experience will be invaluable in steering OpenAI towards our goal of ensuring AGI’s benefits are universally accessible.”
Upon his appointment, General Nakasone shared his enthusiasm for his new role, stating, “OpenAI’s mission resonates deeply with my principles and my background in public service. I am eager to contribute to ensuring that AGI remains a safe, transformative technology that serves global interests.”
General Nakasone’s career has been distinguished by his leadership roles, notably as the longest-serving director of U.S. Cyber Command and the National Security Agency. His responsibilities there involved protecting the nation’s digital infrastructure and enhancing its cyber defense capabilities. His extensive experience spans various high-stakes environments, including cyber units in the United States, the Republic of Korea, Iraq, and Afghanistan, positioning him uniquely to advise OpenAI on navigating the cybersecurity landscape effectively. SOURCE
Conclusion
The recent developments at OpenAI and Zeta Labs showcase the dynamic integration of artificial intelligence in various sectors. With the appointment of Retired U.S. Army General Paul M. Nakasone to OpenAI’s Board of Directors, the company is bolstering its commitment to cybersecurity, ensuring its AI technologies are both safe and beneficial. Meanwhile, Zeta Labs is innovating with Jace, an autonomous web agent that simplifies complex online tasks for users, demonstrating the practical applications of AI in everyday business activities. These advancements underscore the transformative impact of AI, offering both robust security measures and enhanced operational efficiency. For more AI News visit Arcot Group.
