5 ms·

The first comment I saw nailed it for me: in the 80's and 90's there was lots of investment in computers as the PC revolution took off, and the productivity gai

The first comment I saw nailed it for me: in the 80's and 90's there was lots of investment in computers as the PC revolution took off, and the productivity gains were minimal. As with all new technologies, it takes a while for there to be macroeconomic scale changes to industry, and those changes aren't going to come without the up front investment. This article (and frankly, much of society) is looking at the ROI on the wrong time scale.

levocardia 2y ago
Ditto for electrification of factories. It took a decade or more for productivity gains to improve.
- lesuorac 2y ago
  I think that's mostly people have a poor mindset. They think I have this workflow of A->B->C->D and how do I improve "B" using technology. Instead you should have something else of A->E->D where E wasn't possible or maybe just not practical. You don't want to use technology so that you can type keys faster; eliminate the typing of keys process. A lot of the current GenAI is around making images and summarizing text. Those largely aren't real business objectives. If you make a powerpoint 70% faster does that really close a sale 70% faster? I highly doubt it.
  - cbsmith 2y ago
    A lot of the 80's adoption of the PC meant that people did the same presentations but they looked better. ;-) In fairness, it's not just mindset. It takes a while to understand technology and figuring out how it can be effectively harnessed. A->E->D isn't achieved simply by realizing you can do that. There's a whole sequence of steps to change the shape of A,E & D to make that happen. But you're right, a lot of GenAI right now is not being used for things that are real game changers. This is the first step to establishing the foundation from which the real game changers will spring.
  - TylerE 2y ago
    Powerpoints? No. But if it gets good/easy/reliable/cheap enough I can see a lot of applications. Imagine storing incoming customer emails and being able to fish out all emails complaining about, say, slow shipping, even if they don't use that exact phrase.
    - chasd00 2y ago
      > fish out all emails complaining about, say, slow shipping, even if they don't use that exact phrase. isn't that just vectorized or "semantic" searching though?
      - jazzyjackson 2y ago
        honestly fuzzy search in semantic space would be a huge step for most companies to retrieve relevant docs windows is still exact string match but at least allows partial or reordered string match filtering an excel column is exact substring only
  - ilaksh 2y ago
    Funnily enough, I happen to be currently working on an agent application where the goal is to generate a PowerPoint faster. Right now it takes about a week. We should be able to do it in less than 5 minutes. We could do it faster, but I am actually reading and manipulating Excel and PowerPoint examples to make it easier to change the process or adapt the tool calls to other applications, just by editing the instructions and rough templates.

levocardia 2y ago

Ditto for electrification of factories. It took a decade or more for productivity gains to improve.

lesuorac 2y ago
I think that's mostly people have a poor mindset. They think I have this workflow of A->B->C->D and how do I improve "B" using technology. Instead you should have something else of A->E->D where E wasn't possible or maybe just not practical. You don't want to use technology so that you can type keys faster; eliminate the typing of keys process. A lot of the current GenAI is around making images and summarizing text. Those largely aren't real business objectives. If you make a powerpoint 70% faster does that really close a sale 70% faster? I highly doubt it.
- cbsmith 2y ago
  A lot of the 80's adoption of the PC meant that people did the same presentations but they looked better. ;-) In fairness, it's not just mindset. It takes a while to understand technology and figuring out how it can be effectively harnessed. A->E->D isn't achieved simply by realizing you can do that. There's a whole sequence of steps to change the shape of A,E & D to make that happen. But you're right, a lot of GenAI right now is not being used for things that are real game changers. This is the first step to establishing the foundation from which the real game changers will spring.
- TylerE 2y ago
  Powerpoints? No. But if it gets good/easy/reliable/cheap enough I can see a lot of applications. Imagine storing incoming customer emails and being able to fish out all emails complaining about, say, slow shipping, even if they don't use that exact phrase.
  - chasd00 2y ago
    > fish out all emails complaining about, say, slow shipping, even if they don't use that exact phrase. isn't that just vectorized or "semantic" searching though?
    - jazzyjackson 2y ago
      honestly fuzzy search in semantic space would be a huge step for most companies to retrieve relevant docs windows is still exact string match but at least allows partial or reordered string match filtering an excel column is exact substring only
- ilaksh 2y ago
  Funnily enough, I happen to be currently working on an agent application where the goal is to generate a PowerPoint faster. Right now it takes about a week. We should be able to do it in less than 5 minutes. We could do it faster, but I am actually reading and manipulating Excel and PowerPoint examples to make it easier to change the process or adapt the tool calls to other applications, just by editing the instructions and rough templates.