Agentic AI Lacks Human Emotion, Reasoning, and Inhibition

[Joe Leote]

Discussion of Agentic AI with human-in-the-loop:

Why Agentic AI Is Creating Mountains of Slop Code? - YouTube

The technological tools called an AI harness lets Large-Language Models interact directly with computers rather than simply answering questions at a prompt. This means AI can theoretically write production code and execute other tasks without human supervision. The trouble is humans already have enough problems supervising other humans. I am reminded of the line from I-Robot: "Robots building robots, now that's just stupid!"

Joe 

[William Meyer]

Two aspects to ai - the tech itself and the business cycle

I am very much blown away by my interactions with it.  I have asked about very niche topics, not just simple questions but involving complex scenarios and I am just blown away.  I feel like an old percheron on the farm when the row tractor arrived off the flat car.

On the business cycle I would draw a comparison to the railroads.  Most people believe mistakenly that they were overbuilt.   But not competing on a particular line due to competition was not an option if your line was ever to become the next new york central.  Even the winners suffered when the losers went bust but that was only a temporary setback.  Similar story with Amazon and the .Com boom.

--
You received this message because you are subscribed to the Google Groups "Understanding Money" group.
To unsubscribe from this group and stop receiving emails from it, send an email to understandingmo...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/understandingmoney/CAMpwg51HmXfO3tVOH-9gW95cpRiW24Cfeg%2BggcUxzoNRLquaKw%40mail.gmail.com.

[Joe Leote]

The beauty of the government regulating and stabilizing the market system is that the private market over-investments which fuel a technology-producing bubble during overshoot are subsequently allocated as a private market loss during the collapse:

https://youtu.be/rhQx0y89UV4?t=309

Railroads are not the only industry which had investment overshoot and collapse. I think there were over 200 car companies at one time in the USA which consolidated into the Big Three via mergers and/or bankruptcy of the many competitors. In my view the government almost always plays a significant role in the advance of new technology since the Industrial Revolution. Most technology growth stocks have an overshoot and collapse pattern in the long-term price-volume chart. When the stock market crashes the prior advance is then recognized in hindsight as a bubble.

Joe

To view this discussion visit https://groups.google.com/d/msgid/understandingmoney/CABEX-nBtzPTDS59mA5Dc2XP4St3RvxK_q0x1%2BgJ%3Dtu_4ph9M%2BA%40mail.gmail.com.

[Joe Leote]

Frankly I don't understand some of the AI related jargon in this short rant about the business and politics of AI:

Palantir CEO Alex Karp says 'something has gone completely wrong' with how AI is sold

Prometheus is an AI startup. More significantly the character Prometheus (Greek: Forethought) stole fire from the gods. The law uses forethought within the reasonable person standard. The individual, like the ideal reasonable person, is expected to foresee the outcome of actions with a legal duty to take precautions to prevent harm to others.

Based on my admittedly fuzzy or vague ability to decode jargon I would say that Palantir has put the AI harness inside the Ontology layer that somehow limits actions derived from hallucinations and uninhibited impulses of a black box large language model. In other words human engineers have to put an artificial inhibition harness around the agentic AI harness. This requires human forethought, however, the agentic AI systems are supposed to deal with highly complex domains which humans cannot manage all that well via forethought. The talk under the link also discusses the AI business (profit) model coupled to the legal challenges of ownership and control.

Joe