Yes, co-pilot does not go through an excel spreadsheet line by line to count. It truncates. You do need to prompt better. Tell co-pilot the issue you are having with it, along with how "you" can ask co-pilot to give you better results. The limit are 500 rows.
This is a really important point. At the current point in AI development, AI is a tool, like, say, a table saw or welder. Unbelievably powerful but fairly useless in the hands of someone who has no clue as to how to properly use it.
It is not uncommon to see someone complain about their AI experience and their post is so poorly written it's almost unintelligible-- pretty much betraying the fact that the poor performance was almost certainly due to poor prompting.
However, I believe that this "cluelessness gap" will soon be closed (see below).
In post #15 you said AI was junk, just a fast compiler and not a breakthrough. That's where we disagree. It's not junk.
AI has also surpassed the fast compiler phase, because it no longer operates on the strict program rules it is running. That's the difference; that's the breakthrough.
[...]
I would say we are in the AI infancy; in a few years time, and beyond, AI will be on another level.
This is another really important point-- AI is really very young, virtually an infant at this point. It's growing up fast, though, and any limitations today will likely be gone tomorrow. When is "tomorrow?" Difficult question (see below).
And one of the limitations that will disappear (I believe) is the cluelessness gap that I described above. As AI matures, it will almost certainly get better and better at performing properly even in the hands of an incompetent operator.
I put A.I. into much the same category as self driving cars. They're not going to put the body shops out of business anytime soon.
To some degree, I agree with this point. While there are few that are more maximalist in their vision for AI than I am, I also happen to be among the most pessimistic about timelines for AI.
I have no doubt the AI future is coming, that FSD will be perfected, auto accidents will become a thing of the past, and that in the next century people will look back in amazement that their predecessors willingly took part in such a dangerous activity as unaided motoring.
However, I think there is a very high likelihood that the AI believers (of which I am a top-tier member!) are significantly underestimating the long/fat-tail problem of getting the last 1% of AI/FSD figured out. The "march of the 9's" problem is real!
This is my guess: I will be very surprised if AI/FSD is solved to a level of autonomy before 2030, and I will equally be surprised if it is
not solved to that level by 2050.
So, my guess is that you are right in the sense that body shops are unlikely to be out of business anytime soon, but I would caution against concluding that that delay means they will never be out of business.
It may not happen "tomorrow,", but I'm pretty sure that by 2050, auto accidents will only happen to those who take on the risk of driving cars without FSD.
196 entries. 
