lets say I install this extension pangram. it is only scratching
the surface of "verified" information and opens a whole can of worms..
how do i guarantee there is no mitm attack when it makes api calls..
the extension might even assume a mitm attack was a malicious user or attempted jailbreaking
and respond accordingly (filters, etc) but there is a 3rd party..
how do i know there are no mitm prompt injections when it scans the article..
which model is the system using to curate, and it hallucinates or has bias,
hidden system directives. there could be semantic injection..
if you look at the internet landscape as a sigint environment
there can be psyops conducted without dropping leaflets, but direct to my phone/laptop
how do i know, linked-in, social web is not already poisoned with hidden prompts to be ingested
by AI crawlers.. what does this mean for news during elections for example?
if a detector flags content incorrectly at scale, the false positives become
a consensus.. "crowd hallucination"
this is at machine speed, outside human token/s ooda loop
its vital for them to have self validation for exponential rsi.. and this human distillation of human in the loop debugging ai models is needed even though they have judge models handling parallel speculative execution.
labs have parallel speculative execution. they spawn hundreds of agent branches, validate them internally with AI judges and only show the user the successful result.
free users are using sequential single-turn generation. the model requires and waits for the human to debug, fix and re-prompt.
by forcing a human to act as validator. they are capturing high value correction trajectories (Bad Output --> Human fix). They are using your cognitive labour to train judge models and validator agents needed to automate the internal verification step, eventually closing the loop for fully autonomous recursive self-improvement.
human in the loop debugging isn't a bug; it's the necessary training signal for the self-validating agents required for exponential recursive self improvement. With new 'distilled judge' models landing in 2026, this article means that they might have gathered enough data. we might be in the final phase..
For user interface designers. A sincere request to consider these old interfaces like Palm OS, IBM CUA, Mac System 6 for a modern Linux GUI.
The simple black and white interface reduces cognitive load and decision fatigue. Modern UI like skeumorphic, material, drop shadows, 3d interface, aero, glass etc are high cognitive load for some individuals.
A text heavy interface and GUIs that are explicit, not implicit and learnt through discovery is easier for me.
A link to the an article. The picture shows that the IBM CUA works for both terminal and Windows 2.0 type GUI in simple black and white.
From the website;
Optimizations for windowed games improves gaming on your PC by using a new presentation model for DirectX 10 and DirectX 11 games that appear in a window or in a borderless window.
When these optimizations are used, games that originally use the legacy blt-model presentation can use the newer flip-model presentation instead (if the game is compatible). This results in lower frame latency and lets you use other newer gaming features; for example, Auto HDR, and variable refresh rate (for displays that support it).
how do i guarantee there is no mitm attack when it makes api calls.. the extension might even assume a mitm attack was a malicious user or attempted jailbreaking and respond accordingly (filters, etc) but there is a 3rd party..
how do i know there are no mitm prompt injections when it scans the article.. which model is the system using to curate, and it hallucinates or has bias, hidden system directives. there could be semantic injection..
if you look at the internet landscape as a sigint environment there can be psyops conducted without dropping leaflets, but direct to my phone/laptop how do i know, linked-in, social web is not already poisoned with hidden prompts to be ingested by AI crawlers.. what does this mean for news during elections for example?
if a detector flags content incorrectly at scale, the false positives become a consensus.. "crowd hallucination"
this is at machine speed, outside human token/s ooda loop