arxell

This statement that AI researchers are taking up smoking was made by Alexander Karp (CEO of Palantir), who just personally bought 37,000 acres of Swedish forest land, presumably in preparation for some form of AI-related geopolitical fallout. It was absolutely not a peer reviewed study or any sort of survey. It was likely something he heard in passing, a less-than-half-serious joke, or just a subjective questionable observation based on an N of a few. Regardless, this narrative was only picked up and amplified by media because he is the CEO of Palantir. He himself is a cigar smoker. Take that for what you will. I have seen no evidence that it is true -but if it were, I'd imagine there would be other indulgences, vices and financial/credit opportunities much more likely to be observed than smoking.

When every company calls themselves AI-first, what then is an 'AI Company.' In many regards, this hype-chasing branding exercise is currently no different than calling one's company a 'technology company'... Just as the term "AI" itself has (to a degree) become meaningless... and AGI will be soon to follow.

The basis is that this is what we know to be human consciousness is to a degree relateably similar to our own. I do see your point but IMO, a dramatically different form of 'consciousness' comes with the risk of being something completely different from what we consider 'consciousness'. One of the great challenges with speaking about 'Self-awareness' aka consciousness (sometimes incorrectly confused with sentience, sapience, phenomenal experience etc) in AI is that we largely lack the vocabulary to properly capture the mutli-dimensional phenomena that most understand simply as 'consciousness'. IMHO, an ant or dog is 'conscious' in that it can perceive the outside world and make decisions about how to respond to it or operate within it... but we know that we operate at a different level of self-awareness including metacognition. Other forms of 'consciousness' will certainly manifest in future AI systems and perhaps already have but the models and architectures of today lack a number of key features that a human like consciousness would be based upon... a short list might include a persistent, unified self-model across time; continuous learning and development, embodied perception, interaction with the world; intrinsic goals, drives, emotions and deeply integrated metacognition—awareness of its own thoughts, uncertainty, attention, and intentions.

Each has it's pros and cons. Dense models of equivalent total size obviously do run slower if all else is equal, however, the fact is that 35A3B is absolutely not 'a lot smarter'... in fact, if you set aside the slower inference rates, Qwen3.5 27B is arguably more intelligent and reliable. I use both regularly on a Strix Halo system... the Just see the comparison table here: https://huggingface.co/unsloth/Qwen3.6-35B-A3B-GGUF . The problem that you have to acknowledge if running locally (especially for coding tasks) is that your primary bottleneck quickly becomes prompt processing (NOT token generation) and here the differences between dense and MOE are variable and usually negligible.

Totally agree on the consumer and SMB play (which is why we're stealthily working on it :). I'm curious what capabilities the next generation of models (and HW) will provide that doesn't exist now. Considering Ryzen 395 / Digits / etc can achieve 40-50+ T/s on capable mid-size models (e.g., OSS120B/Qwen-Next/GLM Air) with some headroom for STT and a lean TTS, I think now is the time to enter but seems to me the 2 key things that are lacking are 1) reliable low-latency multi-modal streaming voice frameworks for STT+STT and 2) reliable fast and secure UI Computer use (without relying on optional accessibility tags/meta).

My greatest concern for local AI solutions like this is the centrality of email and the obvious security concerns surrounding email auth.