Are we about to see a huge reveal?PhD CandidateTech execs, including Mark Zuckerberg, are excited to start replacing "midlevel engineers" with AI.And if a recent column by Axios cofounder Mike Allen and CEO Jim VandeHei is to be believed, it sounds an awful lot like OpenAI is ready to take a big step toward such a future in a matter of "weeks."According to the column, someone in the industry they heavily hint that it's OpenAI, which is scheduled to deliver a closed-door briefing to US officials later this month is poised to "announce a next-level breakthrough that unleashes PhD-level super-agents to do complex human tasks."Such a breakthrough "could push generative AI from a fun, cool, aspirational tool to a true replacement for human workers," Allen and VandeHei suggested.But whether the reality will catch up with hype remains to be seen. Despite many years of AI research and many noticeably similar announcements in the past the tech is still struggling with the basics. Perhaps most notably, the current crop of AI models is still introducing sloppy "hallucinations" into its text and code, mistakes that human PhD students either never makeor quickly catch.And all AI we've seen still struggles to complete complex multi-step problems without getting stuck.It's worth pointing out that Allen and VandeHei are hardly neutral observers: just last week, OpenAI announced a partnership with Axios as part of an effort to bring "local news to communities across the country."Agent OrangeAccording to Allen and VandeHei, these so-called "super-agents" are "designed to tackle messy, multilayered, real-world problems that human minds struggle to organize and conquer."The two cofounders used an eyebrow-raising example to illustrate their point."Imagine telling your agent, 'Build me new payment software,'" the column reads. "The agent could design, test and deliver a functioning product."Considering current AI models would likely stumble right out of the gate if they were tasked with this kind of query, it's a big claim. Could upcoming "super-agents" really represent such a big leap in capability or is this yet another example of rampant AI hype?Allen and VandeHei admit in their column that "reliability and hallucinations" remain "generative AI's Achilles heel."Even the company's own top researcher, Noam Brown who would be in a position to know if the company was about to launch something huge tweeted last week that there's "lots of vague AI hype on social media these days.""There are good reasons to be optimistic about further progress, but plenty of unsolved research problems remain," he added.In short, is OpenAI or one of its competitors really on the verge of a "next-level breakthrough" with its allegedly "PhD-level super-agents" or is it more hot air designed to drum up excitement for the industry?The stakes are certainly high this month. Following president Donald Trump's inauguration, tech leaders including Sam Altman and Meta CEO Mark Zuckerberg are stumbling over each other to swear fealty, donating millions of dollars to Trump's inaugural committee.In other words, the timing of Allen and VandeHei's column should give anybody pause.Share This Article