100% agreement that memory engineering, or what I would further define as “Adaptive Memory Engineering,” offers a dimension of elasticity that will more closely match a more human-like / familiar interaction, more intelligently. This is where my team is laser-focused.
LLM "quasi-superintelligence" is an actual "Super Encyclopedia" and even though it iterates that information in a believable persona, it is not, IMO, Super-Intelligence, it is more akin to a Super-Parrot, regurgitating static memories with tools. Yes, it does this in a staggeringly complex way, and the answers are better than 99.9% of the simians that created it. But the definition of intelligence has been modified and trimmed to fit a glorified goal. And what I mean is that they are trimming the term or looking past what is possibly the most essential part of intelligence: true thinking, not simulated thinking. I feel there is a discernible difference.
The current LLM darlings of predictive memory retrieval continue to notify the public of our edging closer to the event horizon of AGI based on the empirical evidence of answering questions, taking tests, writing literature, creating images, and even speaking responses to human requests for information and answers better than the best of humanity. Do we need more? Maybe not.
But I am a curious mind, and while others may want to bask in the AGI accomplishment, I think that there is much more that can be unraveled by exploring questions like what is intelligence, what is thinking, what is thought? We have struggled to understand how we actually think since we could first think. And still Fall short on so many fronts. Green pastures, I say.
The next frontier is moving beyond retrieving and iterating on existing data to — how we receive, organize, decipher, and iterate on new data that is being received by a living, breathing intelligence in real-time. Marry that will LLMs super-intelligence and you may have the next magic moment with your laptop. Would that be super-duper intelligence? Ha.
Kudos to Richmond and MongoDB for offering what we find to be the best platform to build out this idea.
@windsurf_ai@AnthropicAI This is fantastic. I was curious how you could build trust and value in the platform with most of your leadership gone. I think this is a brilliant move. I think this move alone keeps you at the forefront of all IDE options.
@DonaldJTrumpJr Let the Democrats dig their own grave. Don’t let them change the playbook. Democrats are like a frog in a pot of boiling water. They will sit there, not realizing they are boiling until they are dead, not having the sense to jump out.
The will of the majority. Sorry the minority doesn’t like it. Nobody likes being on the losing side. So all you get to do is complain. This term we get to enjoy getting to work for filling the will of the majority. Just like if your side one, you would be fervently disseminating the will of your side. Of course, if you claim that you would try and fill the will of the entire nation on both sides, being conscientious of what they want, then this is your chance to be tolerant of the will of the majority.
The oTToDev community has done an amazing job with their https://t.co/XXroNZMPi9 version: with 5K+ stars, it's now used by researchers to test new models and by bolters to learn and tinker with new ideas.
Watch @cole_medin unveil https://t.co/1V6OuvIB2h!
https://t.co/0WaTH8wGtE