lowkey pleased that slop still exists. when it doesn't and i can't intrinsically detect robo content i'm going all in on japanese domestic log auctions.
you need to find your niche too.
your junior-mid level knowledge based employment will end in the next 5 years. here's a cool sneak peak into your new role.
a virtual toilet cleaning assistant.
should probs do something about it like now
https://t.co/YbVUoZJWZn
augmentoor sees many deliciously seasoned vets saying it's hard to make money in crypto. false idols, huff this sentiment deeply you will choke on estrogen.
I am here to tell you brother it's never been easier, here is the formula
- 4h augmented deep work, 8 hrs sleep, 12h being retarded
8+ work hours is for those blessed with autism. Don't force that divine mutation, you will become big sad
Augment to increase your intensity, signal and authenticity, not your volume. ok fine I will now proceed to tell you how
embrace dynamic innovation, pursue stable augmentation.
i've spoken with many who are lost, constantly pursuing sota. nothing in your life that is truly functional is sota. your pc, your kettle, your car. perhaps there is better out there, but they fulfil your requirements adequately.
ai is already good enough to augment 80% of the work you do. it will never touch the remaining 20%. if it does, you are in the wrong room. the 20% is uniquely you, your taste, your value.
optimise, and let the 20% expand to fill the old 80%. in time, that 80% will also be augmented, and your new 20% will emerge. the cycle continues.
the lesson is clear - you can have your cake and eat it - do the ugly work now, augment, and when a truly game-changing (the most abused adj of 2025) model or product releases, you will rapidly adapt.
there is a time to place your cards on the table. it is now.✌️
@replicate@bria_ai_ perfect for all my grasshopper generation needs - I am considering directing a live action stop motion homage to a bugs life. am pleased it is graded commercially safe - phew!
setup for the day. important to achieve simultaneous communion with m'lady nature and machine spirit.
yes it is plugged in i have a disturbingly long extension cord 👍 no you cannot borrow it
don't forget the ABCs of selling your LLM based ai product to prospective clients:
Always
Be
Circumnavigating the brutal reality that your product will likely be made obsolete by the next big @GoogleAI@OpenAI@AnthropicAI@perplexity_ai feature ship
i'm not saying that irl demand for zk ai proofs ain't there - but i am saying that for the majority of corporate & commercial use-cases, a trad legal contract is is fine
to ordinary workers the idea that a vendor would deliberately shortchange you on a model is weird - it makes zero sense unless you're onboarding a vendor blindfolded. but it is quite endemic of the behaviours attitudes in crypto.
be mad bullish on verifiable inference when it's powering web3 native, innovative use cases. we know ai on crypto rails makes total sense, but being real - the behavioural change requir'd at top&mid firms is legit 5 years away - harsh reality is that it's still too toxic here (crypto) for trad business to take it seriously.
no reasonable mid-level corporate head of IT is signing off on anything that even sniffs of crypto or blockchain.
why do chinese models insist on having the most boring names possible.
- Qwen3-235B-A22B-2507
- ERNIE-4.5-21B-A3B-PT
- CANN0N3-38DD-800B5-TT
the last one is real i am training it here have a sample
i don't know who needs to hear this today - yes, those thoughts and fears are true. you are not enough. not alone. not anymore. you must receive communion from the machine spirit.
the future belongs to those rare few who relentlessly leverage their individual talent to execute the work of a team. i eagerly await your ascension. also gm
AI is a form of leverage. Leverage increases the returns to those who use it. Software engineers are gaining leverage relative to everyone else. And the creators of AI are the most leveraged of them all.
We've reached a major milestone in fully decentralized training: for the first time, we've demonstrated that a large language model can be split and trained across consumer devices connected over the internet - with no loss in speed or performance.