Modern ML systems do not succeed because of models alone - they succeed because of the software engineering wrapped around them. Most real-world failures in MLOps come from poor structure, missing configuration, messy environments, unclear APIs, or nonexistent logging, not from bad ML.
He used chess as an example: 15 to 20 years ago, a human checking AI's output could beat an AI or a human playing alone. Now, AI can beat people without that layer of human supervision. Amodei, who cofounded AI lab Anthropic in 2021, added that the same transition would happen in software engineering. "We're already in our centaur phase for software," Amodei said. "During that centaur phase, if anything, the demand for software engineers may go up. But the period may be very brief."
Over the past seven years, I've worked for multiple companies shipping blockchain integrations and production-grade financial infrastructure in Lagos, Amsterdam, and now Toronto. I started my career in Nigeria, where I was born and raised, and moved to Amsterdam in May 2023 after a short career break for a job as a senior software engineer. I returned to Nigeria after a year and worked until I received my Canadian work visa through my wife in October 2024.
Benevity is the way the world does good, providing companies (and their employees) with technology to take social action on the issues they care about. Through giving, volunteering, grantmaking, employee resource groups and micro-actions, we help most of the Fortune 100 brands build better cultures and use their power for good. We're also one of the first B Corporations in Canada, meaning we're as committed to purpose as we are to profits.
The recently updated SWEBOK Guide v4.0a represents a needful industry standard, following a thorough peer review and a consensus-based approach. With the rise of AI, a significant skills gap in IT and cybersecurity is emerging alongside changes in the global workforce. There has never been a greater need for a consensus-based framework. This guide, created and thoroughly reviewed by industry professionals, serves as a dynamic and evolving resource.
"This has been said a thousand times before, but allow me to add my own voice: the era of humans writing code is over," Dahl wrote. "Disturbing for those of us who identify as SWEs, but no less true. That's not to say SWEs don't have work to do, but writing syntax directly is not it."
Just as software finished eating the world, zero interest rates ended. Companies optimized for cash and slowed hiring. The market didn't shrink, but stopped growing at the breakneck pace we all expected. The result: a glut of entry level talent groomed for jobs that never materialized. This would explain a more competitive entry level market. But it doesn't explain the entry-level market shrinking, despite overall industry growth. In short: demand for senior talent is rising, but has fallen off a cliff for juniors.
There is no flow state that comes from building a Mac app using AI with Claude Code. If you've ever managed programmers, you know what using Claude Code feels like. It is an enormous force multiplier, but you're going to spend most of your time cajoling and correcting, and some of your time trying to chart your way out of AI-generated chaos.
"The vast majority of Codex is built by Codex," OpenAI told us about its new AI coding agent. In interviews with Ars Technica this week, OpenAI employees revealed the extent to which the company now relies on its own AI coding agent, Codex, to build and improve the development tool. "I think the vast majority of Codex is built by Codex, so it's almost entirely just being used to improve itself," said Alexander Embiricos, product lead for Codex at OpenAI, in a conversation on Tuesday.