The AI Foundation, which ambitiously hopes to develop “ethical” AI agents that can be trained to complete tasks, today closed a $17 million funding round. A spokesperson said the funds will be used to scale the company’s platform, which allows people to create a digital persona that mirrors their own.
As the pandemic makes virtual meetups a regular occurrence, the concept of personal AI — tailored to your life or replicating your personality — is gaining steam. Startups creating virtual beings, or artificial people powered by AI, have raised more than $320 million in venture capital to date. As my colleague Dean Takahashi points out, these beings are a kind of precursor to the Metaverse — a universe of virtual worlds that are all interconnected, as in novels such as Snow Crash and Ready Player One. Like Soul Machines, Brud, Wave, Samsung-backed STAR Labs, and others, the AI Foundation intends to will the Metaverse into existence by pursuing new and existing machine learning techniques.
Established in 2017, the AI Foundation, which was founded by Rob Meadows and former EA vice president Lars Buttler, is a dual commercial and nonprofit enterprise with the stated mission of bringing “the power and protection of AI to everyone in the world so they can participate fully in the future.” At the center of this mission is the company’s proprietary “human-AI collaboration” platform, which brings together innovations from third-party research partners.
The AI Foundation’s steering committee — the Global AI Council — has a mandate to anticipate the negative effects of AI and figure out how to counteract them. One proposal that reached production is Reality Defender, a free tool that scans images, videos, and other media for known deepfakes and supports the reporting of suspected fakes. Underlining the deepfake threat, Amsterdam-based cybersecurity startup Deeptrace found 14,698 deepfake videos on the internet in June and July, up from 7,964 in December 2019.
Reality Defender runs content through AI-driven analysis techniques to detect signs of alteration or manipulation, while the tool’s community of users flags false positives to retrain the underlying models. It also draws on the AI Foundation’s partnerships with content creators to establish an “honest AI” watermark that clearly identifies AI-generated deepfakes.
Reality Defender is the first of what the AI Foundation calls Guardian AI, a responsibility platform built around the idea that everyone should have personal AI agents working on their behalf. The goal is to devise protections against the risks AI currently poses and to build value for people as AI alters the nature of society.
To further its research, the AI Foundation recently partnered with a team at the Technical University of Munich’s (TUM) Visual Computing Lab to explore new media forgery detection and flagging techniques, leading to the publication of several papers and the release of a public data set that other researchers can build upon. Twitter cofounder Biz Stone joined the AI Foundation last May as director of its nonprofit 503(c) arm committed to releasing tools, tech, policy, and guidelines for shielding against — and mitigating — risk.
Late last year, the AI Foundation hinted at its broader ambitions with Digital Deepak, an AI assistant meant to help users achieve wellness and mindfulness milestones. Built and trained with the AI Foundation’s platform, Digital Deepak, which debuted on the Tonight Show, sounds more or less like Indian-American author and alternative medicine advocate Deepak Chopra. Digital Deepak offers personalized advice and answers questions on “every aspect of well-being,” according to the AI Foundation. “The platform enables everyone to create their own AIs; extensions of the owners that look, sound, and think like them without replacing their humanity … Individuals can fully realize their potential by using their AI to travel to billions of places at once, connect with millions of people, undertake continuous self-improvement, enjoy immersive entertainment, and craft their legacy in the world.”
The AI Foundation’s series B was raised in partnership with Mousse Partners, You & Mr. Jones, Founders Fund, Alpha Edison, and Stone. It brings the San Francisco-based company’s total raised to around $28 million, following a $10.5 million series A in September 2018.