A few days ago, a viral post entitled “Something Big is Happening” was shared on X by Matt Shumer.

It’s a breathless, data-less, fact-less gushing screech that AI is going to come and steel all white collar jobs in the next 12-24 months.  According to Shumer, he’s blown away by what he sees the latest AI models accomplishing.  He wants you to be blown away, too.  This essay was reprinted in Fortune and shared on many news sites.

Is it worth a read?  In a word: no.

When I read Situational Awareness 18 months ago, I sat up and took notice.  Leopold Aschenbrenner’s idea of the intelligence explosion was quite interesting and it was worth considering.  Of course, Aschenbrenner had several things going for him that Shumer doesn’t:

• He worked for OpenAI in frontier research
• He provided copious data and information to support his claims
• He wrote in an academic style about things that have already happened and made logical projections, not hyperbolic End-of-Days statements
• Aschenbrenner hasn’t lied to the world in the past

Matt Shumer is a Fraud

While Shumer’s essay is being reported widely, part of the story is not being told quite as widely: Matt Shumer has lied and committed fraud in the past.  In 2024, he released Reflection, a 70B LLM open source model that he claimed outperformed all of the leading models from OpenAI, Claude, and others.  His web site had an API and when the model was used from that web site, performance was impressive.

But when he put the model on HuggingFace and others tried it, it performed about the same as Meta’s Llama 3, and worse than Llama 3.1.  When people went back and queried the API to compare, they discovered that the model would self-identify as Claude.  It gave the same results as Claude.  And then after a tweak, it refused to use the word Claude.  As it turns out, it was just a wrapper for the Claude API.

Why should anyone ever listen to Matt Shumer again?  Why is he even allowed to work in the field?

Scare-Mongering

Leaving its very questionable author aside, the essay presents the usual prognostications that AI is going to steal all jobs.  It’s the old “if AI is this good now, imagine how good it will be in ____ months!” argument.

To date, none of these AI-induced layoffs have happened.

Shumer makes many unsubstantiated claims in the essay, with no evidence to support them.  While AI is truly a “power suit” for experienced developers, the idea that it will replace all software engineers in the next year or two is not supported by any real-world experience.

Shumer’s main point is that every white collar job – accounting, law, medicine, etc. – is going to be wiped out.  His language is like nuclear war survivalists – husband your resources, save as much as possible, devastation is coming.

…because Matt Shumer says it is.  If Shumer had presented something like Situational Awareness with that level of detail and argumentation, that would be one thing.  But this essay is just his own personal thoughts.  You could grab a random college student, hand them them a bong, and ask them to speculate and you’d get something with the same intellectual rigor.

How Will Shumer Say He’s Sorry?

Inevitably, some people read Shumer’s article and are now scared.  Some high school senior is deciding to change his major.  Some family that is scraping by is terrified that their breadwinners are going to be out of jobs.  This is Matt Shumer’s fault.

How will he say he’s sorry?  Promising to leave the field and never say anything publicly again might be a good start.

raindog308

Raindog308 is a longtime LowEndTalk community administrator, technical writer, and self-described techno polymath. With deep roots in the *nix world, he has a passion for systems both modern and vintage, ranging from Unix, Perl, Python, and Golang to shell scripting and mainframe-era operating systems like MVS. He’s equally comfortable with relational database systems, having spent years working with Oracle, PostgreSQL, and MySQL.

As an avid user of LowEndBox providers, Raindog runs an empire of LEBs, from tiny boxes for VPNs, to mid-sized instances for application hosting, and heavyweight servers for data storage and complex databases. He brings both technical rigor and real-world experience to every piece he writes.

Beyond the command line, Raindog is a lover of German Shepherds, high-quality knives, target shooting, theology, tabletop RPGs, and hiking in deep, quiet forests.

His goal with every article is to help users, from beginners to seasoned sysadmins, get more value, performance, and enjoyment out of their infrastructure.

You can find him daily in the forums at LowEndTalk under the handle @raindog308.