Gyroscopic Investing

Xan wrote: ↑Fri Feb 17, 2023 1:47 pm

yankees60 wrote: ↑Fri Feb 17, 2023 1:29 pm No Laughing Matter

But this is!

https://www.penny-arcade.com/comic/2023 ... -must-bing

(hey, looking at the hyperlink I see that must be a reference to the Harlan Ellison story!)

btw:

https://twitter.com/thetrough/status/16 ... 98020?s=20

Latest Southpark Episode. Featuring ChatGPT.

Mark Leavy wrote: ↑Sat Mar 11, 2023 1:35 am Latest Southpark Episode. Featuring ChatGPT.

The b____ a___ in Sw______ !

Mark Leavy wrote: ↑Sat Mar 11, 2023 1:35 am Latest Southpark Episode. Featuring ChatGPT.

I LOL'd.

But also, does ChatGPT have good game? Might need to explore that angle...

Kriegsspiel wrote: ↑Sat Mar 11, 2023 9:03 pm But also, does ChatGPT have good game? Might need to explore that angle...

I don't play in the same crowd as you, Kriegs. But a few times I've gotten a long rambling email from the girlfriend and I couldn't make heads nor tails of it. I asked the AI to break it into relevant bullet points, summarize the points and suggest a response. I usually didn't use the response part, but the rest was extremely helpful. Think of it as an AI translator that is fluent in both male and female speak.

Mark Leavy wrote: ↑Sat Mar 11, 2023 9:14 pm a long rambling email from the girlfriend

Now I’m laughing (in sympathy)

Version 4 is scarily smart. I don’t have access. I’m just going by the Cinderella example on the front page. wow.

Mark Leavy wrote: ↑Sat Mar 11, 2023 9:14 pm But a few times I've gotten a long rambling email from the girlfriend and I couldn't make heads nor tails of it. I asked the AI to break it into relevant bullet points, summarize the points and suggest a response.

Absolutely classic!

dualstow wrote: ↑Thu Mar 23, 2023 7:31 am Version 4 is scarily smart. I don’t have access. I’m just going by the Cinderella example on the front page. wow.

Yeah I thought the test scores were interesting:

710/800 on SAT reading+ writing
700/800 SAT Math

https://openai.com/research/gpt-4

4/5 and 5/5 on the AP exams. I believe there is a writing component graded by humans on those exams.

This is moving at incredible speed. Just this week ChatGPT opened up to plugins. There is already a list of familiar names that have plugins ready to go, but most every major tech company will have a ChatGPT plugin in short order. Plus any user can write their own plugin to enable ChatGPT to do pretty much whatever you want.
I can't see how this would not eliminate millions of jobs. Not the least of which will be consulting jobs. But I guess it will also escalate human efficiency by orders of magnitude. A friend of a friend recently wrote a full blown business plan in 3 days using ChatGPT that otherwise would have taken him months to do.

For my part, I'd like to see if ChatGPT can create a 1985 Kelly LeBrock.

Jack Jones wrote: ↑Fri Mar 24, 2023 6:11 am

dualstow wrote: ↑Thu Mar 23, 2023 7:31 am Version 4 is scarily smart. I don’t have access. I’m just going by the Cinderella example on the front page. wow.

Yeah I thought the test scores were interesting:
...4/5 and 5/5 on the AP exams. I believe there is a writing component graded by humans on those exams.

Impressive

glennds wrote: ↑Fri Mar 24, 2023 10:30 am For my part, I'd like to see if ChatGPT can create a 1985 Kelly LeBrock.

I don't like it when people say "you win the Internet today" so I'll just say,
This is the finest line I will read all weekend.

glennds wrote: ↑Fri Mar 24, 2023 10:30 am


For my part, I'd like to see if ChatGPT can create a 1985 Kelly LeBrock.

No 1985 Steven Seagal also?

yankees60 wrote: ↑Fri Mar 24, 2023 3:48 pm

glennds wrote: ↑Fri Mar 24, 2023 10:30 am

For my part, I'd like to see if ChatGPT can create a 1985 Kelly LeBrock.

No 1985 Steven Seagal also?

Why not?

Garage in, garbage out. Believe Wiki? (For those who are enthralled by it as an always credible source.)

Mountaineer wrote: ↑Sat Mar 25, 2023 4:56 pm

yankees60 wrote: ↑Fri Mar 24, 2023 3:48 pm

glennds wrote: ↑Fri Mar 24, 2023 10:30 am


For my part, I'd like to see if ChatGPT can create a 1985 Kelly LeBrock.

No 1985 Steven Seagal also?

Why not?

Garage in, garbage out. Believe Wiki? (For those who are enthralled by it as an always credible source.)

Not sure what the last section (highlighted) means or its relevance. But for many years ALL Seagal movies were must see movies who enjoyed that movie genre.

ChatGPT just totally flubbed the knights & knaves logic puzzle.
I still love it, though. Maybe version 4 does better.

dualstow wrote: ↑Mon Apr 03, 2023 5:37 pm ChatGPT just totally flubbed the knights & knaves logic puzzle.
I still love it, though. Maybe version 4 does better.

So far it's 3 for 3 on giving me good recipes. Two nights ago I asked it for a good recipe for roasted Greek lemon potatoes. I used the recipe it gave me with no changes and it was outstanding. Before I would have googled it and ended up reading 3-5 recipes and choosing one. So now I just trust Chat GPT for the recipe I'm seeking.

So I wanted to try something more eccentric. I asked it for instructions on how to cut a corner dovetail joint in wood by hand, and the instructions it gave me were pretty good. Of course, cutting dovetails well takes practice, but all the steps and correct sequence was there.

That makes sense. It can read and regurgitate instructions pretty well, sometimes very well.
It can translate fairly well.
It just doesn’t do a lot of thinking.

I pasted an article to it today that I’d already read and it summarized it very well. It also correctly answered questions I asked about the article.

Another view of ChatGPT:

Subject: ChatGPT has quite the imagination



Just for fun – because I’ve written about the subject and I wanted to see

how much of my work would be used – I asked ChatGPT to “write a history of

pitchers not batting ninth in major league baseball.” I guess the bot

didn’t find my piece, but I’m not sure how hard it looked, because

virtually every sentence in the resulting copy was false, bat-guano crazy

false, and I don’t think there’s any source for it anywhere, which makes me

wonder how in the world ChatGPT came up with this stuff. Here’s some of

what it produced:



*“This tradition [batting the pitcher ninth] held firm for decades, with

few exceptions. One notable exception occurred in the 1930s, when St. Louis

Cardinals manager Frankie Frisch occasionally batted his pitchers eighth in

the lineup, in order to take advantage of the better hitting abilities of

his position players. This strategy was later dubbed the "Frisch Rule,"

although it was not widely adopted by other managers at the time.”*



Frisch never, in his 2200+ games as a major league manager, put his pitcher

anywhere but ninth in the lineup. The “Frisch Rule”? Where did THAT come

from? And what it in the world is “in order to take advantage of the better

hitting abilities of his position players” supposed to mean?



*“It wasn't until the 1970s that another manager challenged the traditional

lineup configuration. Chicago White Sox skipper Chuck Tanner experimented

with batting his pitchers in various spots in the order, including eighth

and even seventh. However, these experiments were short-lived, and Tanner

eventually returned to the traditional ninth spot for his pitchers.”*



Tanner never, in his 2000+ games as a major league manager in games with no

DH, put his pitcher anywhere but ninth in the lineup. In the entire decade

of the 1970s, for all 24/26 teams, the pitcher batted somewhere other than

ninth three times.



*“**The next significant challenge to the pitcher batting ninth tradition

came in the 1980s, when Whitey Herzog managed the St. Louis Cardinals.

Herzog's teams were known for their speed and aggressive baserunning, and

he believed that batting his pitcher eighth in the lineup would allow his

position players to take advantage of more scoring opportunities. This

strategy was known as the "Whiteyball" style of play and was quite

successful, as the Cardinals won the World Series in 1982. Other managers

soon followed Herzog's lead, and by the mid-1980s, several teams were

experimenting with batting their pitchers eighth.”*



This is the craziest passage in the whole piece. Herzog never, in the 1500+

National League games he managed, put his pitcher anywhere but ninth in the

lineup. Furthermore, no pitcher for any team batted anywhere but ninth in

the entire decade of the 1980s. “Whiteyball”? Batting his pitcher eighth

“would allow his position players to take advantage of more scoring

opportunities”? An eighth-grader faking a book report could come up with

more plausible stuff than this.



*“In recent years, there have been occasional experiments with batting the

pitcher in other spots in the order, such as seventh or even sixth.

However, these experiments have been short-lived…”*



And of course, recent years – especially the period from 2015-2019, and

Tony LaRussa with the Cardinals before that – have seen the pitcher bat

somewhere other than ninth far more than any other period in major league

history (at least post-1900, I haven’t looked at the 19th century). But

since 1973, the pitcher has hit seventh only twice (both of them Dontrelle

Willis, 2005) and never sixth (the last pitcher to do that was Gary Peters

in 1968).



I know ChatGPT doesn’t always get things right, but to be this wrong takes

some work.

A.I. cant even keep spam out of my inbox.

Looks like someone else wrote that, Vinny. email fwd?

dualstow wrote: ↑Wed Apr 19, 2023 9:21 am
Looks like someone else wrote that, Vinny. email fwd?

Yes, of course. I thought it might have been fairly obvious but I guess it was not?

So your entire signature is built on lies!

dualstow wrote: ↑Wed Apr 19, 2023 10:17 am
So your entire signature is built on lies!

I did "provide" what someone had written in an email.

Who?

dualstow wrote: ↑Wed Apr 19, 2023 2:05 pmWho?

It was in here: SABR-L Digest - 11 Apr 2023 to 12 Apr 2023 (#2023-74)

A research place for baseball related items.