So much for Artificial Intelligence...

[rmhinden]

His avoidance of remembering things was based on his theory that his brain had a finite ability to remember stuff, and if he used up his memory on non-consequential stuff when he was young, he would not be able to remember anything at all when he got old.

I think your co-worker is wrong. Our brains are dynamic and create new pathways based on new stimulus. He is also confusing short term vs. long term memory.

I think he is more likely to loose memory by not excersing his brain.

Bob

 

[smorgasbord]

His avoidance of remembering things was based on his theory that his brain had a finite ability to remember stuff, and if he used up his memory on non-consequential stuff when he was young, he would not be able to remember anything at all when he got old.

A Conan Doyle wrote very similar thoughts from Sherlock Holmes:

https://archive.li/HcqjF

Sherlock Holmes would not have been surprised to hear that human memory is limited in capacity (''When the Brain's Mailbox Is Full,'' July 27). Tasked by Dr. Watson for his ignorance of the Copernican theory of the solar system in ''A Study in Scarlet,'' Holmes said he considered that a man's brain was ''originally like a little empty attic'' and that it was ''a mistake to think that that little room has elastic walls.'' There comes a time, Holmes adds, ''when for every addition of knowledge you forget something you knew before.''
Later, in ''The Five Orange Pips,'' Holmes expanded on what he originally told Watson: ''I say now, as I said then, that a man should keep his little brain-attic stocked with all the furniture that he is likely to use, and the rest he can put away in the lumber-room of his library, where he can get it if he wants it.''
Fortunately I had the lumber-room of the Web to find these half-remembered bits of Holmesian arcana.

 

[luvmytoolz]

I think your co-worker is wrong. Our brains are dynamic and create new pathways based on new stimulus. He is also confusing short term vs. long term memory.

I think he is more likely to loose memory by not excersing his brain.

Bob

I may forget what I walked into the kitchen for, but I can name the song and artist of most 80's hits just from a note! ;-)

 

[WillAdams]

Computers today are quite fast, and with multiple cores, able to process in parallel --- with the commensurate power consumption, hence the building of data centers for processing AI queries.

 

[Packard]

I may forget what I walked into the kitchen for, but I can name the song and artist of most 80's hits just from a note! ;-)

There is some science behind that forgetfulness. Walk back to the location where you first had that thought and you will likely have an “Ah, yes. Now I remember” moment.

The mind tends to want to “close the door” on thoughts as you leave an environment. You can probably find out why, if you can think how to pose that question to A.I. without forcing it into some cyclical thought maze.

 

[luvmytoolz]

There is some science behind that forgetfulness. Walk back to the location where you first had that thought and you will likely have an “Ah, yes. Now I remember” moment.

The mind tends to want to “close the door” on thoughts as you leave an environment. You can probably find out why, if you can think how to pose that question to A.I. without forcing it into some cyclical thought maze.

It's funny you say that, I first heard the reasoning behind not trying to force a memory years ago.

We can be watching a film and just can't quite remember the name of an actor we know well, so I deliberately push it from my mind and don't think about it, and almost always within a minute or three, the name suddenly pops into my head!

 

[Packard]

It's funny you say that, I first heard the reasoning behind not trying to force a memory years ago.

We can be watching a film and just can't quite remember the name of an actor we know well, so I deliberately push it from my mind and don't think about it, and almost always within a minute or three, the name suddenly pops into my head!

I asked A.I. the right question. It’s called the “doorway effect” and it is apparently a real thing. This from the BBC:

Although these errors can be embarrassing, they are also common. It’s known as the “Doorway Effect”, and it reveals some important features of how our minds are organised. Understanding this might help us appreciate those temporary moments of forgetfulness as more than just an annoyance (although they will still be annoying).

Why does walking through doorways make us forget?

Forgetting why you entered a room is called the “Doorway Effect”, and it may reveal as much about the strengths of human memory, as it does the weaknesses.

www.bbc.com

 

[Cheese]

Here's another Google AI faux pas...they got the write-up correct but mistakenly attached a video of the ETSC 2 to it. That seems to me to be a larger sin because most folks will watch the video and believe the video over the written explanation.
To further confuse everyone, the photo with the green arrow is a picture of the DTSC 400.

So, the original query/discussion is about the DTSC 200 while the photo is a DTSC 400 and the video is an ETSC 2. AI at its best.

 

[Packard]

Here's another Google AI faux pas...they got the write-up correct but mistakenly attached a video of the ETSC 2 to it. That seems to me to be a larger sin because most folks will watch the video and believe the video over the written explanation.
To further confuse everyone, the photo with the green arrow is a picture of the DTSC 400.

So, the original query/discussion is about the DTSC 200 while the photo is a DTSC 400 and the video is an ETSC 2. AI at its best.

Interesting, but seems insignificant compared to the damage A.I. is capable of. It has been blamed for a suicide, for planning an anti-Semitic assault, for loss of a database and other irreversible errors.

10 famous AI disasters

Insights from data and ML algorithms can be invaluable, but be warned — mistakes can be irreversible. These recent high-profile AI blunders illustrate the damage done when things don’t go according to plan.

www.cio.com

Additionally, the Internet is rife with similar stories. Here is my search:

Google Search

​

 

[smorgasbord]

There's a great podcast from The Center for Humane Technology that dives into a specific instance where ChatGPT ended up encouraging a teen to take his own life.

TL,DL:
Almost all AI chat boxes are designed to make the user continue the chat. That means some combination of reinforcing what the user is telling the AI with suggestions on continuing that line of thought. While ChatGPT initially did try to steer the teen to suicide hotlines and other resources, the teen got around that by saying he wasn't asking about suicide for himself, but only as an exercise to get data for a book he was writing. But, when be complained that he didn't feel comfortable talking to his family, the chatbox said something along the lines of "you can always talk to me and I won't tell anyone."

It's a sad story, and to me shows that the effects of Unintended Consequences are real and potentially devastating.

Also interesting to hear that in China, most AI is not for chatboxes like ChatGPT, but to solve business problems, as in manufacturing optimization.

 

[Coen]

Also interesting to hear that in China, most AI is not for chatboxes like ChatGPT, but to solve business problems, as in manufacturing optimization.

What China promotes on their platforms for western consumption is very different than what they promote domestically.

 

[ChuckS]

While AI is used extensively in commercial, industrial and defense sectors in China, it is also widely employed by the young generation like ChatGPT in the west, such as DeepSeek Chat https://www.deepseek.com/en/

Google: DeepSeek is an advanced, cost-effective AI chatbot from China, similar to ChatGPT and Claude, designed for conversational interaction, complex reasoning, coding, and mathematical problem-solving. It offers specialized modes like "Deep Think" for in-depth analysis and supports file uploads for analysis.

One main difference between AI Chat tools in China and ours is that they're heavily monitored by the authority, just like its internet activities (in which many western apps are banned and blocked). You can't, for example, expect any meaningful responses from its AI apps to any questions considered "sensitive" by the authority.

In addition to the suicidal tragedy (he showed his neck bruise from his first failed attempt to his mom) and other negative outcomes, MIT researchers have just found, subject to peer review, that AI could result in what is termed as cognitive debt for young learners. AI is a double-edged sword.

 

[Joelm]

My wife just told me that by adding "-ai" to your Google search it won't give you an AI answer and just behave like it used to.
Going to start adding -ai from now on.

 

[luvmytoolz]

My wife just told me that by adding "-ai" to your Google search it won't give you an AI answer and just behave like it used to.
Going to start adding -ai from now on.

Not just on Google, boolean searches also work great on sites like Ebay, to help filter out specific classes of items.

 

[Mini Me]

Google has had for many years an advanced search page that is totally different to the normal one and it is based around Boolean searching in a very simple way that enables it to be used by anyone.

Google Advanced Search

 

[mino]

Google has had for many years an advanced search page that is totally different to the normal one and it is based around Boolean searching in a very simple way that enables it to be used by anyone.

Google Advanced Search

Unfortunately, their search engine no longer respects your query ... for about 10 years, last five it is worst, G is basically an ad engine today. Getting it to provide relevant results is a tall, tall order.

Yes, my GooFu was well known ... not anymore as I resigned to waste time trying to fight the Ad engine which is heuristically /read unpredictably/ ignoring the logical parameters to "do its thing" or literally just throws out /false/ "no results" responses when you are too hard on it.

I still use Google, when all else fails, but unlike in the past, I have learned the hard way it cannot be trusted especially when "no results" is the response. In a way, we are back to 2000 when searching was a hit-miss game and bookmark collections ruled the game.

What China promotes on their platforms for western consumption is very different than what they promote domestically.

Indeed. And that is quite logical given how the West treated /and treats/ Huawei. Or how ASML acts. We act hostile, they adapt. As expected.

They do not even try to market their cloud services in the West - given there is no point in our unpredictable regulatory environment. What they do offer is mostly commercial on-premise stuff that is "easier" to sell into an irregularly regulated market. This creates the false impression of Chinese ICT companies being behind on LLM use and deployment while nothing can be further from the reality.

What is more interesting is the Russian take. They are pretty strong on AI in general but they are not all that much into LLMs, instead focusing on the other, not so compute-heavy branches of AI research and use. Will be interesting where this approach leads them five years down the line.