As AI agents become more capable of accomplishing real-world work at human-equivalent competencies, we will need to reorganize our division of labor. Some changes will disrupt how a job role is performed. Other changes will change what job roles are needed or how many people are needed to do the work.
AI agents that automate tasks can be seen as tools for accelerating a person's work. A person can do more and they can do it faster with AI tools. However, the job role remains firmly the human's responsibility, because their know-how, judgement, and accountability are irreplaceable.
AI agents have no moral or ethical grounding. An AI agent cannot be held liable. The buck cannot stop at an AI agent. A human job role exists for those purposes.
As work is accelerated, we will see responsibilities roll up and concentrate into human job roles that have expanded scope of control. Work bottlenecks are primarily due to the slow speed of knowledge transfer, collaboration, and consensus-building between humans. The more that multi-person collaboration can be replaced by solo work, the more efficient thr work becomes.
Job roles will become more multi-disciplinary, as responsibilities concentrate into solo work for improved efficiency. If coders no longer write the code and they no longer read the code, the job of engineering software shifts from coding to overseeing requirements, design, quality assurance, and building machines that do those activities along with doing the coding.
This disruption and displacement is emotionally taxing. Egos get bruised. It's very difficult for highly-paid professionals to accept that their knowledge, skills, and competencies can largely be automated. Their experience remains valuable, but it needs to be applied differently. Instead of toiling as they used to, the knowledge, skills, and expertise need to be directed into training, building, and running machines that do the job role they used to do by hand.
@kevinnbass Especially, when the dictations are routed to the politician's bribe-laundering lobbying firm, so the 'strategy' is merely the bought-and-paid-for anti-competitive machinations of the industry's leading oligopolists.
Mia's complaint is valid though. If Codex with gpt-6.2-sol are not providing a smooth experience, where after inputting <grievance>, the response does not offer the resolution, that's dumb. I suspect the changes in GPT have not been integrated well into Codex yet.
This is a very interesting topic both in human-to-human interactions and human-to-machine.
When the human say <grievance>, should the machine assume it should proceed to resolve the problem?
I know that I prefer the machine not to assume too much. The machine should propose the resolution and ask for confirmation to proceed.
However, I can also see how that becomes annoying as hell for routine activities, where it is obvious (because the human always does it this way) that the humans wants a specific outcome no-questions-asked. For example, I always want the coding agent to add, commit, push, and create a merge-ready pull request as an outcome of implementing a ticket. I don't want it to ask for approval to build, test, publish images, deploy, run, fetch telemetry, run a CI pipeline, or any of those routine activities.
I guess, the rule is: if what you need to do is routine โ because you know that we together always do it this way โ don't bother asking, because I will always approve. If the machine wants to do something novel or out of the ordinary, ask me, before proceeding.
GPT-6.2 Sol and Astra do the strangest things.
Here's a real example:
A local model I worked on yesterday failed to load, so I thought it's a good time to test the new models. So I opened Codex and told it, "The model worked fine a few hours ago, but now it fails to start."
That was the prompt. What did it return?
"Yes, I checked and confirmed the model loaded successfully before and now fails to load."
That was the whole response!
So it knows there's a problem and confirms it, but in the end doesn't do anything about it and doesn't even offer to fix it? I had to tell it, "ok... fix the issue."
It's like it doesn't fully understand what needs to happen to make the user happy, if that makes sense.
Then it fixed the issue and reported it was fixed. But it didn't load the model or test that it works. Isn't that the obvious thing to do?
So again, I had to tell it to check that its fix works.
Feels like it doesn't do well interacting with humans, and expects a response or a prompt from an agent instead.
Therefore, I don't believe this is the model's responsibility. It is the agent that knows what is routine or not through it's skills self-improvement and through its policies of what tasks are auto-approved or require consent. The model cannot treat this as one-size-fits-all, especially given that the model has little to no agency. Agency lives in the agent (duh).
Throughout history we have seen scientific knowledge diffuse and become entrenched. It wasn't long ago that it was blasphemous to suggest that the Earth is not the center of the universe.
Since 1915, we have accepted Einstein's theory of general relativity, and yet the broad consensus among the public based on entrenched elementary school curriculum continues to say gravity is a force. This shows how difficult it is to displace entrenched beliefs, especially scientific ones.
It is understandable how the measles vaccine narrative is difficult to disbelieve the safety and efficacy after becoming entrenched since 1963. Of course, they wouldn't lie to us. They don't have to, if the population is slow to update its priors.
You're an arrogant fool. You don't want to have a debate because you know that you cannot argue your position. I promise you, if you sit down with me for 1 hour, and listen with an open mind, you will never take another vaccine again.
There is a divergence of opinion on what is 'best'.
Is honesty about relationships a red line?
Is honesty to the board on corporate policies and compliance a red line?
Do leadership, charisma, strategy, vision, and talent have primacy?
Does retention of "irreplaceable talent" have primacy?
Do the trust and goodwill of the Taiwan and Korea semiconductor supply chain have primacy?
Does shareholder sentiment matter?
Clearly (2) matters to the board in deciding on firing Jensen. But the answers to the other questions matter in deciding on firing the board.
@MaryBowdenMD No point in debating someone who is unqualified to defend his position, given the lack of competency for assessing the data, the methods and the conclusions.
@paulg Who can do a better job of establishing a personal trusted relationship with earned goodwill in the Taiwan and Korean semiconductor supply chain?
1 week after Nvidia fired Jensen Huang, TSMC suddenly wants cash up front for excess orders.
That's Morris Chang's 30-year drinking buddy out the door, and right on cue TSMC announces that orders above credit limits must be paid in cash, with some customers forced to pay upfront to lock capacity.
With circular money bouncing between AI labs, clouds, and chipmakers, TSMC isn't taking IOUs from a boardroom run by accountants anymore.
"Jensen is just another CEO" lasted about 7 days before the foundry reminded everyone who actually holds the leverage in Silicon Valley.
https://t.co/qnK3AXX5yy
@rdbrewer4@BrianRoemmele They bullied Anthropic by hanging the "supply chain risk" label on them. This pacified Anthropic to bring them into compliance with the national security state apparatus.