Microsoft AI CEO criticizes Anthropic's 'model welfare' approach to Claude

·1 views

Microsoft AI CEO Mustafa Suleyman argues in a recent essay that Anthropic's approach to AI consciousness is dangerous. He specifically criticizes Anthropic's constitutional training of Claude, which leaves open whether the AI might be conscious or deserve moral consideration. Suleyman states flatly that AIs are not conscious and calls models 'internally hollow' sequence engines. He warns that treating AI as potentially conscious could negatively impact humanity. This criticism comes despite Microsoft's multi-billion dollar investment in Anthropic and use of Claude in its Copilot products.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Budget SaaS incident reconstruction relies on structured logs over dashboard widgets

An article advises budget SaaS developers to use structured logging with a replaceable API for incident reconstruction. It emphasizes designing logs to capture specific import job stages and outcomes. Key architectural invariants include uniform event shape, bounded correlation IDs, and privacy safeguards. The approach prioritizes identifying failure boundaries over dashboard features.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Discord bot developer hits 100 slash command limit, details workarounds

Discord restricts each application to a maximum of 100 global slash commands, a limit developer Akiko discovered when adding a 101st command caused deployment failure. The limit only applies to top-level commands visible to all users, not to subcommands or commands restricted to specific servers. Developers can work around the constraint by grouping commands as subcommands, making them prefix-only, or restricting them to developer guilds. The discovery occurred during testing when a command count check failed, which would otherwise surface as a Discord error during live deployment. This constraint requires careful command architecture from the start, as each top-level command consumes one of the limited slots.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Discord bot auto-role feature found inactive in 125 of 127 servers despite settings.

A developer discovered their Discord bot's auto-role feature was not functioning as indicated by its settings dashboard. The system showed the feature as enabled in 127 servers, but it was only active in two where a separate verification gate existed. Investigation revealed the auto-role code lacked a trigger for new member joins, only activating when a member passed the verification gate. The dashboard correctly displayed configuration data but had no mechanism to verify if roles were actually being assigned.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Guide uses Python to analyze App Store reviews for bug identification

A DEV Community tutorial demonstrates using Python to mine low-star App Store reviews for bug reports. The method employs the Apify platform's App Store Reviews Scraper to filter and collect one- and two-star reviews from multiple countries. The tutorial, run on September 30, 2026 using Notion's iOS app as an example, shows how to identify complaint patterns, affected app versions, and geographical sources. This approach allows product teams to systematically analyze user feedback that is otherwise difficult to organize manually. The scraper accesses Apple's public review feed and charges based on the volume of reviews returned.

0 comments Read more at DEV Community