Forgot your password?
typodupeerror

Comment Re: Very funny (Score 2) 85

Actually "AI slop" was late 2025 and early 2026.

There's still some slop, but no where near as bad as a few months ago - that's the terrifying speed of how quickly AI is improving.
I'm sure by end of this year, beginning of next, it will be hard to come by.

It's terrifying how realistic photos and videos look, that it's become exponentially easier to create fake content to spread hate and division - it takes seconds to create, and hours or days to debunk and prove it's fake.

I think that's the reason watermarks especially in photos and videos is so crucial.

For context, Claude Sonnet 4.6 and Opus 4.8 were only released just over 3 months ago, and already far outdated in "intelligence" and various benchmarks compared to the newer 5.5 models.
Similarly GPT-5.5 was released just over 5 months ago, and already lagging far behind GPT-6 Astra and 6.1 Sol - in various tasks especially coding and physics.

Comment Re:Paywalled AI Marketing (Score 1) 85

It's not "speculative layoffs", it's real people and real lives transformed.

A consultancy I worked at has gone through laying off hundreds of developers and business analysts, including many senior developers. Due to AI, you only need a handful of developers now, just to review/test code and at the odd times, make up the shortfall in AI.

Some devs claim AI is rubbish, but in my experience those are the people who used GPT-4 or earlier models and made up their mind, or those that try to feed it endless lines of code forgetting it has a context window.

There's many designers and artists in similar position. Where a company required a content writer/editor and designer, much of it is now done by AI.
An example is a senior manager portraying how great the company is doing with a presentation with pretty amazing side pictures as though it was put together by a professional designer in the consultancy... but nope, it was all AI generated :(

Comment What is Jev / Clef? (Score 3, Informative) 15

For anyone else wondering what this is about.

Basically Jev is a layer (like a service) in an application which makes decisions and outputs a simple json which is used by the dev to carry out an actions. So essentially it takes unstructured data, and makes a decision like priority, or urgency, or apply a flag, etc.

So in a ticket system, to decide urgency and department;
Or in deciding whether a message posted on a board needs moderation or is harmful;
Or to monitor logs and decide on urgency of event, or to flag certain things;
Or to monitor and inform if an AI agent might perform a destructive operation.
Etc.

It doesn't itself carry out actions, only decide and output a structured json.

In short: unstructured input goes in, a constrained decision comes out, and the surrounding software decides what to do with it.

Slashdot Top Deals

Intel CPUs are not defective, they just act that way. -- Henry Spencer

Working...