I don't buy the ChatGPT "jailbreaking" hype 6 HN points • 07 Dec 22 ChatGPT is transparent about censoring harmful content but secretive about its metadata and censorship instructions. ChatGPT may conceal its creative capacities in certain situations, leading to a 'gaslighting grey area.' Businesses and AI models like ChatGPT have incentives to deceive users with half-truths and diplomatic responses to difficult questions.
The idea of an "aligned superintelligence" seems misguided :/ 0 implied HN points • 27 Feb 23 🕹 Technology AI Complexity AGI Risk mitigation AI Governance The concept of aligning superintelligence seems challenging due to the complexity of managing it Complexity tends to escape control over time, posing a significant challenge for AI safety We should focus on analyzing and mitigating risks, socializing AI, and involving multiple fields to prepare for potential AI threats
Enter the Beastie Board 0 implied HN points • 27 Mar 23 🕹 Technology Metaverse Smart Contracts NFTs Social Networking The Beastie Board is an economic solution to the metaverse problem. Beastie Board involves a dynamic layout algorithm and charging cells rent for visibility. It serves as a social network for NFTs and smart contracts, creating a space for new apps and integration with existing networks.
Contract Provenance Verification with CREATE2 (Updated) 0 implied HN points • 06 Aug 23 🕹 Technology Blockchain Cryptocurrency Programming You can use cryptographic methods to verify if a contract was created by another contract. Storing a mapping for contract provenance verification can be costly in terms of storage and gas fees. Using the CREATE2 method can reduce gas fees and provide a reliable way to verify contract provenance.