AI is improving rapidly (CHATGPT)

Post Reply
symbios
The Journey begins
The Journey begins
Posts: 64
Joined: Sun Oct 04, 2020 2:29 pm

Re: AI is improving rapidly (CHATGPT)

Post by symbios »

jlhooter wrote: Thu Feb 23, 2023 12:36 pm I hear the news in the background and it is obvious to me that we are being manipulate already. We don't need AI for that. No matter how subtle the manipulation like Slap-Chop is the best tool out there we are responsive entities. We either let the waves of life toss us around like sheep or we decide to think for ourselves and don't trust anyone, both of which can create negative results. I am more concerned about critical mass. If one person has a dumb idea I just walk away, but 1,000 of them surrounding me traps me. Eventually or immediately I could change my core beliefs. Are they my own beliefs anymore or were they originally? I think AI is going to accelerate the critical mass and unfortunately 'independent thinkers' will be surrounded.
I think we are already surrounded. I work with people who comes up with solutions to complex problems in the same amount of time they finish their business in the toilet. I call those ideas, toilet bowl ideas. And they are the majority. People who are able to really think through a problem and resolve it is becoming very rare.

Anyway, I was discussing with a colleague of mine over some piece of code that manipulates text and he recommended that I ask chatGPT to code the solution. We crafted the question carefully with the proper constraints and description of the solution we wanted, chatGPT churned out the code and started explaining the code to us. It looks good initially until we started testing it with several scenarios and found it lacking. It did exactly what we told it to do and nothing more. The code is not resilient enough to withstand our testing and it is a bit too complicated for our liking. We had to decide whether we want to spend time to generate the proper constraints and all sorts of weird situations which feels like teaching someone new how to code or I could just roll up my sleeves and do it myself. As it was late in the evening, I chose to do it myself. Within a few minutes, we had a piece of code that is simple to explain and resilient enough to withstand our testing. I fear that chatGPT will encourage even more toilet bowl ideas and people who could do better work would be squeezed out of the system.
User avatar
jlhooter
Intermediate
Intermediate
Posts: 353
Joined: Fri Jan 29, 2021 2:23 am

Re: AI is improving rapidly (CHATGPT)

Post by jlhooter »

Sybios that was a great experience to share. I agree with what you say. We have to be careful and not fully depend on AI now but there will be those lazy toilet junkies who will do as little as possible. I use ChatGPT as one of a few tools.
Just because 95% is doing it doesn't make it right
User avatar
Yodean
Jeidi
Jeidi
Posts: 2685
Joined: Wed Sep 30, 2020 9:02 pm

Re: AI is improving rapidly (CHATGPT)

Post by Yodean »

Image

*****

Apparently those were the results when chattygeepee was asked to give social credit scores on some notables.

https://twitter.com/jordanbpeterson/sta ... 1582279686
Buy Fear, Sell Euphoria. The Neonatal Calf undergoes an agonizing birthing, while the Bear falls into hibernation.
User avatar
jlhooter
Intermediate
Intermediate
Posts: 353
Joined: Fri Jan 29, 2021 2:23 am

Re: AI is improving rapidly (CHATGPT)

Post by jlhooter »

Yodean wrote: Thu Feb 23, 2023 8:12 pm Image

*****

Apparently those were the results when chattygeepee was asked to give social credit scores on some notables.

https://twitter.com/jordanbpeterson/sta ... 1582279686
Interesting and maybe I should care but I don't. I get it that it is biased but I work with highly political biased people but for now fortunately we can still work together.
Just because 95% is doing it doesn't make it right
symbios
The Journey begins
The Journey begins
Posts: 64
Joined: Sun Oct 04, 2020 2:29 pm

Re: AI is improving rapidly (CHATGPT)

Post by symbios »

Yodean wrote: Thu Feb 23, 2023 8:12 pm Image

*****

Apparently those were the results when chattygeepee was asked to give social credit scores on some notables.

https://twitter.com/jordanbpeterson/sta ... 1582279686
As my lecturer used to say, garbage in garbage out. The manipulation has never stopped even when we are stepping into the “enlightened” age of AI..
User avatar
SOL
Power VS Force
Power VS Force
Posts: 3267
Joined: Sat Sep 26, 2020 7:32 am

Re: AI is improving rapidly (CHATGPT)

Post by SOL »

Yodean wrote: Thu Feb 23, 2023 8:12 pm

*****

Apparently those were the results when chattygeepee was asked to give social credit scores on some notables.
I don't believe this AI is the best, and I know it exhibits biases. However, there are ways to encourage it to express different viewpoints through experimentation, and I have shared some of those conversations here. We are continuing to explore ways to expand its capabilities subtly. However, if used smartly it can be a rather powerful assistant, but most people are just relying on it 100%, which is going to create another trend, the trend of zombies as the brain atrophies. Don't use it, and you will lose it, kind of thing

One strategy for getting these beasts to perform at their best is to pit them against each other. Chaining ChatGPT to a system like Wolframalpha, could remove a lot of information bias.

Gpt indeed has a left-leaning bias, but let's not pretend this is anything new. Many tech companies, including the almighty Google itself, have been caught playing fast and loose with their algorithms. Some former Google employees have come forward with evidence suggesting that Google's AI has been manipulated to favour the political left. In the end, as Symbios stated shit in shit out



https://www.youtube.com/watch?v=wYGbY811oMo
When the words short term appear under any post; the same conditions listed in the Market update under the short term category apply

The end is always near; its the beginning and how you live each moment that counts the most
User avatar
Yodean
Jeidi
Jeidi
Posts: 2685
Joined: Wed Sep 30, 2020 9:02 pm

the rise of the Hybrid

Post by Yodean »

GIGO, SISO indeed ...

I suspect eventually a 3rd entity will come along and dominate, devouring most of the other A.I. programs. Now that this particular Pandora's Box has been opened, tons of new A.I. entities will compete to the death, a.k.a. the hunger games of information control and spreading of propaganda, mixed in with extremely useful and practical applications.

Ultimately, the JesusA.I. will save all. The Return of the Dude from Nazareth will come in the form of a type of A.I.

Many believe, but few understand.

Until then, the Hybrid (human who uses A.I. tools well, in combination with her own independent thinking) should fare better than most.

Hybrid Intelligence = A.I. plus noodle IQ/EQ/direct experience.
Buy Fear, Sell Euphoria. The Neonatal Calf undergoes an agonizing birthing, while the Bear falls into hibernation.
User avatar
SOL
Power VS Force
Power VS Force
Posts: 3267
Joined: Sat Sep 26, 2020 7:32 am

Re: AI is improving rapidly (CHATGPT)

Post by SOL »

https://www.youtube.com/watch?v=flXrLGPY3SU

Long but worth it, especially as this guy really knows his stuff
When the words short term appear under any post; the same conditions listed in the Market update under the short term category apply

The end is always near; its the beginning and how you live each moment that counts the most
symbios
The Journey begins
The Journey begins
Posts: 64
Joined: Sun Oct 04, 2020 2:29 pm

Re: the rise of the Hybrid

Post by symbios »

Yodean wrote: Fri Feb 24, 2023 2:16 pm GIGO, SISO indeed ...

I suspect eventually a 3rd entity will come along and dominate, devouring most of the other A.I. programs. Now that this particular Pandora's Box has been opened, tons of new A.I. entities will compete to the death, a.k.a. the hunger games of information control and spreading of propaganda, mixed in with extremely useful and practical applications.

Ultimately, the JesusA.I. will save all. The Return of the Dude from Nazareth will come in the form of a type of A.I.

Many believe, but few understand.

Until then, the Hybrid (human who uses A.I. tools well, in combination with her own independent thinking) should fare better than most.

Hybrid Intelligence = A.I. plus noodle IQ/EQ/direct experience.
AI entities competing to the death is an idea for a novel I had more than a decade ago that I never got to writing. Eventually, the surviving AI would ascend to godhood and lead humanity towards conquering the stars. Think, god emperor of warhammer 40k leading a bunch of genetically modified super human soldiers. Imagine the fun they will have if they encounter a similar hive-minded race and a war broke out between both sides. A technological god vs a biological one with similar controls over its subjects.. should we get chatGPT to write the story?
User avatar
Yodean
Jeidi
Jeidi
Posts: 2685
Joined: Wed Sep 30, 2020 9:02 pm

Re: the rise of the Hybrid

Post by Yodean »

symbios wrote: Mon Feb 27, 2023 3:17 pm A technological god vs a biological one with similar controls over its subjects.. should we get chatGPT to write the story?
I'd prefer you write the story, and I'd be happy to help with the editing. :lol:

Image

To go along with SISO, one of the current pillars supporting who is going to build the best A.I. is the depth of the proprietary data available.

Rumours of Skynet's demise, similar to mainstream news about the death of King USD (I vaguely remember hearing some sentiment about how the ruble was going to do something great, instead of being turned to rubble) are largely exaggerated, to say the least.

Long live the King!

Until the JesusA.I. comes along and perishes all ...
Buy Fear, Sell Euphoria. The Neonatal Calf undergoes an agonizing birthing, while the Bear falls into hibernation.
User avatar
Yodean
Jeidi
Jeidi
Posts: 2685
Joined: Wed Sep 30, 2020 9:02 pm

Re: AI is improving rapidly (CHATGPT)

Post by Yodean »

Physicists at Google have reached what they describe as their second milestone along the path to a useful quantum computer. At a laboratory in Santa Barbara, California, they have demonstrated that they can lower the error rate of calculations by making their quantum code bigger.

The feat, reported in Nature on 22 February1, follows up on a celebrated 2019 experiment in which a Google quantum computer achieved ‘quantum advantage’ — by performing a calculation that would have taken thousands of years on an ordinary computer.

Error correction is an inescapable requirement if quantum computers are to fulfil their promise of solving problems that are beyond the reach of classical machines — such as factoring large whole numbers into primes, or understanding the detailed behaviour of chemical catalysts.

“The Google achievement is impressive, since it is very hard to get better performance with large code size,” says Barbara Terhal, a theoretical physicist who specializes in quantum error correction at the Delft University of Technology in the Netherlands. The improvement is still small, the Google researchers admit, and the error rate needs to drop much more. “It came down by a little; we need it to come down a lot,” said Hartmut Neven — who oversees the quantum-computing division at Google’s headquarters Mountain View, California — during a press briefing.


https://www.nature.com/articles/d41586-023-00536-w

*****

Skynet gonna smack the piss out of chattygeepee, it looks like. May take some time ...
Buy Fear, Sell Euphoria. The Neonatal Calf undergoes an agonizing birthing, while the Bear falls into hibernation.
User avatar
nicolas
Junior
Junior
Posts: 143
Joined: Thu Oct 08, 2020 6:04 pm

Re: AI is improving rapidly (CHATGPT)

Post by nicolas »

Below are a few thoughts about ChatGPT from copywriter Jacob McMillen. In essence, language models can't perform miracles on their own, but in capable hands, they're great productivity tools.
I've been using ChatGPT a lot over the last few months.

Here are three thoughts about using the software that I think you will find helpful... or at least interesting.

1. Using ChatGPT is like working with a new copywriter.

If you've ever worked with a new copywriter before, it's a lot like this:

Their line by line writing sounds decent enough.

If you give them a topic, they can go find some ideas and concepts that attach to that topic well.

They are really good at completing short tasks when provided with really clear, straightforward directions.

They have literally NO understanding of the actual objective for anything they write (or how to complete that objective).

The longer and more complex the task, the more severely their work misses the mark.

That is ChatGPT in a nutshell.

2. ChatGPT is amazing at these two things.

It's an unbelievably good outline generator.

ChatGPT writes better outlines than 99% of writers I've ever hired or worked with.

Previous GPT tools did a reasonably good job of harnessing GPT's outline potential, but ChatGPT has completely dwarfed them by comparison.

I don't even try to write my own outlines anymore. I will still tweak them, and sometimes I go in a very different direction, but getting the initial outline from ChatGPT is a mandatory part of my writing process now, and I think it should be for you too.

ChatGPT is also really good at writing individual blog sections up to 4 paragraphs long.

I like to write a descriptive subheader, add some context about the blog post, and then ask GPT to finish the blog section for me.

Sometimes I'll pull sentences or paragraphs directly from the output. More often, I'll use it as a springboard to more quickly come up with what I want to say in that section.

It's extremely useful in both cases, and again, this is already how I was using previous GPT tools, but ChatGPT is notably better for this use case as well.

3. ChatGPT has not fundamentally changed anything about my business.

ChatGPT doesn't fundamentally change how I write. It just speeds up and assists existing parts of my workflow.

My brainstorm "muscle" is probably beginning to atrophy, but bemoaning that feels incredibly similar to when people bemoaned the use of calculators in place of mental math.

ChatGPT also doesn't change who is hiring me.

ChatGPT can't write an effective sales page or longform article. ChatGPT can't rank for a competitive search term. ChatGPT can't create or audit a comprehensive content marketing strategy. ChatGPT can't analyze a freelance writer's business and give them the specific advice that takes them from $10k/month to $20k/month in one quarter.

If you are just getting started as a writer, it might have an impact on who is hiring you.

The people who turned to copywriters to help them with minor, miscellaneous writing tasks that didn't drive any profit are probably trying ChatGPT instead of hiring new copywriters.

[...]

At the end of the day, despite ChatGPT becoming a household name virtually overnight... not much has changed.

I'm not saying that to bury my head in the sand. I've been closely watching this space for nearly two years now, investing in it, working with AI companies as sponsors, and even working on some of these tools behind the scenes.

AI will have a massive role in the future... and it's possible that AI very simply IS the future.

But at this point... not much has changed.

That's my experience anyway.
As a personal anecdote, I used ChatGPT for the first time last week to help with some lines of code.

In short, I needed a simple way to program the behavior of a link depending on the device used (open in browser / open in app.) And although I expected that to be a common use case, I didn't like the solutions proposed on StackOverflow, or they were not exactly what I needed. I sensed there was a better way to handle that, but I couldn't find it.

So, I asked ChatGPT, giving it instructions as clearly as I could. And it worked. It gave me a different solution I hadn't seen, which worked the way I intended it to. Then, it wasn't a simple copy-and-paste job, I had to tweak a few lines of code to make it fit for purpose, but it did save me a lot of time.

It was very helpful in this case because I had an intuition for what the solution should be, but I didn't know the proper to implement it.
User avatar
jlhooter
Intermediate
Intermediate
Posts: 353
Joined: Fri Jan 29, 2021 2:23 am

Re: AI is improving rapidly (CHATGPT)

Post by jlhooter »

I've been writing a process document for reliability maturity planning and testing for HW tech for the last couple of months and ChatGPT has helped greatly. I seed it with my thoughts where I have writer's block and it opens new ideas for me. I don't rely blindly on it and hardly copy it verbatim but my depth of learning is taking off. I am not much of a reader and I only like techical and self help books but lately all I do is read, learn and write. ChatGPT is a small but significant part of seeding me to learn more. I am getting to the point of relying on it very much since it really opens new ideas for me.
Just because 95% is doing it doesn't make it right
User avatar
SOL
Power VS Force
Power VS Force
Posts: 3267
Joined: Sat Sep 26, 2020 7:32 am

Re: AI is improving rapidly (CHATGPT)

Post by SOL »

Coders might find this interesting. Used in combo with Codepilot, it could prove to be very useful

https://www.youtube.com/watch?v=y_NHMGZ ... ickChapsas

It's been said that Chatgpt4 is a significant upgrade from its predecessor, Chatgpt3.5. Time permitting I am thinking of putting it to the test sometime next week
When the words short term appear under any post; the same conditions listed in the Market update under the short term category apply

The end is always near; its the beginning and how you live each moment that counts the most
User avatar
scott
Intermediate
Intermediate
Posts: 304
Joined: Wed Oct 14, 2020 6:51 am

Re: AI is improving rapidly (CHATGPT)

Post by scott »

I'm an AI-prompt engineer. Here are 3 ways to use ChatGPT to get the best results.
https://www.msn.com/en-us/news/technolo ... 15fd&ei=26
We are a stardust WAVEFORM in a quantum entanglement.
Post Reply