orbitalventures

7 วันที่ผ่านมาdiscuss

The issue that I take on this is ego, we believe that we can do a better job sometimes, an we exhaust ourselves with no reason, Agent delegation works in the same way as Human delegation, the difference is the speed on how that happens, reviews were a good idea when humans toke days to do a product, features, code, document or what ever. The reward piece I believe is faster to see something working, the issue seems to be the control we had. The approach needs to be the same of those we do with almost everything else. We drive a car but no necessarily knows how to assemble, we might now here and there trouble shooting, change tires, oil, do some level of maintenance, but at the end just a few of us can pull an engine apart and check every piece, cars are not even manufactured manually anymore, QA is part of an automated process, as production rates are far to fast for humans to intervene. AI brought that to us, insanely fast automation, and with that, Humans should not become gatekeepers but rather guides.

orbitalventures

7 วันที่ผ่านมาdiscuss

LLM's discussion is always the same, if you don't use it you lose it, although true, the question is why and maybe we don't need to do that thing.

Humans have been gravitating into cognitive thinking and not specialized tasks, some of us, a handful do repetitive tasks and that is ok, but, the majority of us work better in context, in guidance. LLM offers that, offers that abstraction that we were and are looking into all aspects of our lives. We once plow the fields, and harvest by hand, and was good, but suddenly became cumbersome, too much to handle and machines came along to do the job of 20, faster and better, and there humans just tweak the machine, teach the small repetitive tasks. We did that with everything and we change our lives, the way that we upskill ourselves and our communities, we let specialized human tasks as a comfort/skill/romantic/artistic outlet, like some specialized vineyards or specialized farms that do specific very specific things. But the rest we delegate and move to another areas. I believe the problem is not LLM/AI is how we are going to up skill, repetitive jobs no longer need us, so we need to upskill for context, for management, strategy

orbitalventures

11 วันที่ผ่านมาdiscuss

Is really anyone surprise about this? To be honest a company that shifts goal posts constantly, that has Government as the only backer, that inflates revenue and talks about Base in Mars and the expansion of human consciousness is the definition of Red Flag. The bond so quickly after the IPO is just another scheme of the kind of people this is being targeted to. As son as you dig slightly deeper into the BizPlan and numbers you realize how much of the entire SpaceX IPO and Bond are scams.

orbitalventures

11 วันที่ผ่านมาdiscuss

Sad sad times indeed, specially because AI empowers more those who seem to do wrong more than good. What we discovered is that implementing a double agent verification, specially in aging population is the way to mitigate. That is family members or more exposed members are required to have a double authorization notification for performing transactions. Banks are doing it quite well and is completely align with MFA. Of course this requires configuration of a trust circle among every person, and generates another avenues for abuse like was pointed. Sadly we are always running behind into protecting people.

orbitalventures

11 วันที่ผ่านมาdiscuss

Implementing LLM Gateways and Policy Enforcers (PE) is the only way to contain against these attacks. We discover that we needed to use PE across all our interactions client or developer facing. We also had to create Enforcer policies in Lite llm versions and re-enforce with ML for threat detection. The approach substantially reduce the amount of data leak, and errors.

My advise for all the ones that are looking into commercial AI applications, USE a PE and an LLM Gateway, do not let your clients reach LLM directly without checking it first.

orbitalventures

11 วันที่ผ่านมาdiscuss

I Agree 100% with this approach, we are discovering that discrete languages are exceptionally good for LLM to deal with them, classic verbs and recommended simple sentences, we have been having quite a good time using it as replacement of Terraform for managing infrastructure and the results are quite promising.

comments