Quote:
Originally Posted by goneil2024
AI - It’s early, and we have a long way to go with AI.
DIGITAL - When I lived in MA, I had a very close personal friend that was in product development/sales at Digital, about 1989-1990. At the time my IBM PC-XT from 1985 was running an Intel 8088 CPU, and my friend gave me a sneak preview of a new DEC laptop “prototype” with a 40MB hard drive, 128K RAM and an Intel 80-486 chip, it even had a dial up modem and weighed 6lb. His comment to me at the time was that “the average consumer could never use the power of that laptop”, and the concern was sales wouldn’t meet expectations. Today I have a LENOVO with Intel Core i9, 16 (6/8/2) cores, 64GB RAM, 2TB M.2 Memory, OLED display, weighs 3+ lb, and all the other bells and whistles. So, don’t underestimate the customer or the ability to adapt (unlike DEC).
PATIENCE - After over 45 years using both mainframes and PC’s/Laptops I am resolved in the belief that “you cant be too rich, too thin, or have too much computing power”. I expect AI will find a place and most users will learn how to leverage the technology. If not then it will fade away, and we will need to navigate the transition. It’s a tool, like the wheel, movable type, etc. and its up to the user to maximize it’s use.
|
I have very similar experiences, living in MA, starting using a DEC mainframe PDP 8i in 1972, Dartmouth Basic timesharing in 1978, MAC in 1982 with Multiplan, the predecessor to MS Excel, trying to use statistics to win the lottery, trying to program a PC in 1986 for automated voyage reports, had a personal PC built in 1988 during grad school 'cause IBM selling PCs in suits was ludicrous, started with Dell towers in 1994/5 building financial models. . started working at an internet LAN/WAN R&D and manufacturing company when the internet was 3 years old, and during the next 5 years, the biggest users of internet at the time was porn watching late at night. (seriously).
in 1995, the engineers were designing ethernet for cable modems for home delivery, and was working with another developer for internet based real estate marketing and sales. . .I never have had enough speed: in 1999, waiting 20 minutes to save changes to the international budget models. . 3 changes per hour was painful. .
However, if you look at the time line from development above to what is almost ubiquitous today, there is longer than a 5 year full adoption cycle for universality. That is where we are today, at the beginning of a 5 + year cycle for software and data which may upset the world.
The fact that ChatGPT public release is trained on cheap(free) data, which is full of human biased crap, is the beginning like porn usage for the internet in the mid 1990s.
Today's data center build is like the fiber optic cable laying in the 1990s, infrastructure preparation prior to the software/usage/customers being developed. . However the ROI is not positive for more than 5 years, which makes the investments look frivolous, and will cause some investment pricing corrections, towards a more conservative, practical rate of investment and rollout as the software gets refined and the use cases start to produce real investment returns. . .
good luck to us
as we worked during the best years of the technology age,
just as our parents lived and worked in the best years of the industrial age, and our grandparents lived at the best but last years of the agrarian age