Big Tech Is Finally Paying for the AI Disaster It Created

Noodles

The sequel will probably be better.
Joined
Sep 20, 2018
Messages
6,260
Location
Illinois
SL Rez
2006
Joined SLU
04-28-2010
SLU Posts
6947
There's no evidence that anyone in the upper leadership of Cisco has any sort of hedge for surviving the AI implosion that I'm increasingly convinced is coming, and seems to be coming soon.

I'm retiring in the next few years, and a significant portion of my retirement funds are tied up in Cisco stock. Am I about to get wiped out? Things that keep me up at night...
From what I can tell, a lot of companies are also pivoting to more "white box" solutions away from expensive more proprietary systems like Cisco has. Which will hurt them even more when the bubble bursts I think.
 

GoblinCampFollower

Well-known member
Joined
Sep 20, 2018
Messages
5,680
SL Rez
2007
I'm retiring in the next few years, and a significant portion of my retirement funds are tied up in Cisco stock. Am I about to get wiped out? Things that keep me up at night...
I have some stock in my company too. That's largely normal, but you definitely don't want it to be a huge portion of your portfolio. So yes.... if you can, diversify a lot of it NOW.
 
  • 2Agree
Reactions: Archer and Beebo Brink

Beebo Brink

Climate Apocalypse Alarmist
Joined
Sep 20, 2018
Messages
7,185
SL Rez
2006
ETA: It's ironic, that at the same time we're being pushed onto the AI everywhere bandwagon "Everyone must be using AI daily!" we got a message Tuesday that was essentially "Hey, maybe you all can use cheaper models?" It seems that token expense is becoming an issue at Cisco.... colour me surprised...
The inability to track the cost of token usage is one of the biggest liabilities the AI companies aren't addressing. It's also going to make the Chinese model -- which is substantially cheaper -- much more attractive.
 
  • 1Agree
Reactions: GoblinCampFollower

Noodles

The sequel will probably be better.
Joined
Sep 20, 2018
Messages
6,260
Location
Illinois
SL Rez
2006
Joined SLU
04-28-2010
SLU Posts
6947
The inability to track the cost of token usage is one of the biggest liabilities the AI companies aren't addressing. It's also going to make the Chinese model -- which is substantially cheaper -- much more attractive.
Yeah, its really annoying that "tokens" are basically like an obscure in game currency device, except you never know if your micro transaction will cost you 20 cents or 20 dollars.
 

GoblinCampFollower

Well-known member
Joined
Sep 20, 2018
Messages
5,680
SL Rez
2007
You would really find this next video interesting. During this Tech Report interview, El emphasizes this very point: the technology can do lots of things, but NOT all the things the techbros are claiming. The gap between the hype and the reality is not just misleading, it's downright dangerous.

Took me a while to get around to watching the whole thing, but it's great. She's of course right that the actually useful things the AI's can do gets buried in all the hype.

To add to her examples, one major concern of mine is that it's become very popular for leaders at my company (and likely many others) to use LLM's to summarizing large amounts of email, documentation etc. to save them time and stay informed. ...but there is a huge problem with this. LLM's are actually really bad at writing good summaries. Sure, they seem to be good at it. They can turn a huge wall of text into a nice, neat, little blurb that is way more digestible, but there are huge fundamental problems with the actual quality of the information. LLM's just fundamentally lack the ability to know if a particular piece of information is important, or just something added in for flavor.

One example I saw a while ago was when I said something to my boss in a weekly summary about how "Job X ran despite major system issues that the team had to do XYZ to mitigate." And his AI summary reduced this to just "Job X ran." ....so the summary lost all of the actual information that even made it worth mentioning.

In short, I think these AI's have often not been extremely helpful to individual contributors to do much of the real work, but the way senior leaders use it is very different and way more troublesome. It seems to really help them FEEL more informed and advised but the information is often really bad. Leadership seems to like the tools a lot more than we peons do, and it's really to the detriment of all.

What I said about summaries is important, but I also know some people who work under leaders who just have full on AI psychosis. They trust the machine god for EVERYTHING and ONLY trust the machine.
 
  • 1Like
Reactions: Archer

Veritable Quandry

Specializing in derails and train wrecks.
Joined
Sep 19, 2018
Messages
5,400
Location
Columbus, OH
SL Rez
2010
Joined SLU
20something
SLU Posts
42
Not only do the miss context and nuance, but LLM summaries have a tendency to actually reverse the main point. It may in fact tell you the job ran when it actually failed.
 

Lori Claremont

Active member
Joined
Sep 20, 2018
Messages
104
Location
Calgary, Alberta, Canada
I have some stock in my company too. That's largely normal, but you definitely don't want it to be a huge portion of your portfolio. So yes.... if you can, diversify a lot of it NOW.
We're definitely considering this. Not sure when the right time to pull the trigger is: Next week is the earnings announcement and it's expected to be quite good, the bubble hasn't burst yet after all....
 
  • 1Like
Reactions: GoblinCampFollower

Argent Stonecutter

Emergency Mustelid Hologram
Joined
Sep 20, 2018
Messages
7,685
Location
Coonspiracy Central, Noonkkot
SL Rez
2005
Joined SLU
Sep 2009
SLU Posts
20780
It seems like I'd agree with her on most of this. A lot of people lament that we need to shut it down, but I think very few actually want to ban the technology or research into it.
Raises paw.

There are so many cases throughout history of useful technology being pioneered by evil entities that cast a dark shadow over it for a bit. A few quick examples that come to mind are Nazi's making jet engines and rockets. Or Monsanto developing so much farming technology.
Except this is more like NFTs or Bitcoin except centralized.
 

GoblinCampFollower

Well-known member
Joined
Sep 20, 2018
Messages
5,680
SL Rez
2007
We're definitely considering this. Not sure when the right time to pull the trigger is: Next week is the earnings announcement and it's expected to be quite good, the bubble hasn't burst yet after all....
Makes sense! You might have to wait till after earnings anyway as an "insider." I'm not even suggesting you get rid of it ALL, just diversify a bit. But obviously I'm not here to run your life.
 

Lori Claremont

Active member
Joined
Sep 20, 2018
Messages
104
Location
Calgary, Alberta, Canada
Makes sense! You might have to wait till after earnings anyway as an "insider." I'm not even suggesting you get rid of it ALL, just diversify a bit. But obviously I'm not here to run your life.
No worries, it's almost like you've been listening in on our conversations! We do have a reasonable 401k and RRSP (the fun of working in two countries over your career ;-) ), but nervous about our exposure from being in a single stock that's at risk...
 

GoblinCampFollower

Well-known member
Joined
Sep 20, 2018
Messages
5,680
SL Rez
2007

Got a new video!

Just more detail into the financial con artists. This one focusing on Palantir. I know it's a long shot, but I am very hopeful that once this bubble burns down, we get a new era of sensible regulations on the stock market shenanigans.
 

GoblinCampFollower

Well-known member
Joined
Sep 20, 2018
Messages
5,680
SL Rez
2007
Agreed. She's very watchable. A lot of what I've thought about watching these videos and many others is that we DO need some safety regulations around the technology itself. ....but that pales in comparison to the stock market abuses we need to reign in. My concerns about AI are dwarfed by how worried I am about circular financing and a number of other really scammy things these tech bros are doing with their stocks such as very skewed voting rights of share holders.