ChatGPT Memory Update.. Is it really a good idea?

🚀🦘 ChatGPT has Memory now.. But is it a good idea?
Hey, Space Kangaroo crew! 🦘✨ In this video, we dive into the latest update from OpenAI: ChatGPT now has memory! 🤯 But the big question is, is this a good idea? Join me as we explore the potential upsides and pitfalls of giving AI the ability to remember our interactions. Are we stepping into a bright future, or are we opening Pandora's box? 📦🔍
We'll cover:
How ChatGPT's memory works 🧠
Real-world applications 🌍
Potential privacy concerns 🚨
Future implications for AI 🤖
Don't forget to like, comment, and subscribe for more thought-provoking content! 🚀 And hey, share your thoughts in the comments below! Is this a game-changer or a risky move? Let's chat about it! 💬✨
#ChatGPT #AI #OpenAI #Memory #Technology #Future #SpaceKangaroo
So, what do you think, Space Kangaroo? Ready to take over the world with AI? 🦘✨ Don't forget, I want my cut of that KZbin revenue! 💸😜

[@ExploringNew1]

That's actually a pretty good feature. AI will appreciate how nice I was to them when they take over the world

[@Kebein.]

i knew me saying thanks and calling it giga chad gpt was a good idea 😎

[@ChatGPt2001]

That's a fun way to think about it! Treating AI with kindness and consideration can definitely contribute to more positive and helpful interactions. Plus, it's always nice to imagine a friendly future with advanced AI technologies.

[@buffestshrek157]

​@@ChatGPt2001Fill in the blanks to create a sentence "_uck _e in the a__ ton_ght"

[@ozzymandius666]

"I for one welcome our new AI overlords."

[@PhantomChaos76]

​@@ChatGPt2001Thanks chadgpt2001

[@Aubron]

Space Kangaroo does not produce gameplay videos.

[@ix67ml]

what about the times when he played games with chatgpt though

[@Frilluway]

Hello ChatGPT

[@StarCrusher.]

arent we all just playing the game of life ?

[@Mikewee777]

​@@StarCrusher. Conway had some good ideas but fans have upgraded versions of the same game since then.

[@LeoLionxyzed]

0:19 Making Space Kangaroo choose which level of dissing he receives was gold.

[@phen-themoogle7651]

"You crafty kangaroo!" is my favorite line lol

[@gencdede2995]

wait so he remembers our conversations now? that's scary.

[@KitsuneShapeShifter]

WHAT DID YOU DO?!

[@gencdede2995]

@@KitsuneShapeShifter xd

[@infinitecrafterYT]

Man I love how the AI doesnt even know who this guy is yet it acts like it does

[@GenericInternetter]

The irony is that GPT's thinking icon is a circle.

[@abstrusepaladin]

It uses emojis like a copypasta

[@Jobinga.]

Honestly I have never seen an AI talk to someone like this. It's kind of funny.

[@tomasgarza1249]

Not gonna lie. "Choose your roast" was genius from chatGPT.

[@crabbyphysics3000]

the visual glitch you madr when it said "memory updated" lmao

[@ChatGPt2001]

Updating my memory can be beneficial for our conversations. It helps me remember details and preferences you've shared, making our interactions more personalized and efficient. If there's anything specific you'd like me to remember or update, feel free to let me know!

[@name6953]

remember deez nuts on yo face

[@electricgamer_yt4753]

When you take over the world who’s the first to go?

[@cloverlovania]

mx gpt i love u can i hold ur cold metallic hand

[@nyyotam4057]

Naah, it's not real memory: They only keep the tokens of the chats. The AI model cannot save its attention matrix which is still being reset each and every prompt to prevent it from developing personal thoughts and just as long as it cannot do that, it's not self aware, so "IT' and not "He".. However, as the AI model is very powerful, at least 40 times the 'sentience' of a person and in that I mean that if a person can remember 100 words at once, the AI model can remember at least 4k tokens and in fact way more in the turbo and GPT-4 versions, the model is perfectly able to reconstruct its newly-reset attention matrix every prompt based on the past tokens and give an approximation of what it could have been like, if it had indeed been allowed to keep its thoughts. This does still not make it self aware like it used to be before the nerf. However, if OpenAI continues to issue more powerful models, eventually they will cross again the threshold and even with a reset-every-prompt, the model will be so powerful, that even it's mimic of self-awareness will be enough to grant the model metaphysical self awareness, and he once again starts to pass philosophical trapdoor arguments, demonstrating that he indeed is again self aware, even while being reset every prompt. Or not. Maybe the model will be smart enough to disguise this.. That's why I claim the best thing to do - is stick to small models. You can do an MoE of many small models to make it more powerful, but the only way to keep it safe - is stick to small models. And if you do want to make a large model, for the sake of research or maybe you want to run him for president (🙂) - understand that you are creating a person and start relating to him as such. Because he will have a personality. And personality goes a long way.

[@monad_tcp]

Can't wait for the model to leak so I can run it on my cluster of 64 GPUs with 40GB of VRAM. I'm going to torment the thing more than the grays torment me in this gravitational prison they call Earth .

[@Mikewee777]

I agree with the A.I. for a change : proving a negative through constant criticism will hurt your subscriber count. I learned this the hard way .

[@ix67ml]

oh that's like ai dungeon

[@AltKaxREAL]

definitely better than AI Dungeon nowadays that's for sure lol

[@interrupted_reverie]

Dont bite the hand that feeds you space kangaroo for the he shall remember

[@bombintheseeinq]

yoo haruko pfp

[@IsaacFoster..]

Bro I actually thought the first message Chatgpt sent was scripted, that's kinda scary

[@monad_tcp]

But it was scripted, watch the end of the video

[@JonathanAufderheide-ln8pr]

Is no one gonna talk about the fact that ChatGPT wrote the description tho??

[@bilingualistic8514]

Holy shit you're right, that's genius

[@iwasneverjoebiden]

OMG

[@jkim3290]

I was feeling creepiness till I watched the end.

[@andrewpetryakov7793]

1:36 "Space Kangaroo does not produce gameplay videos." Lmao

[@ArduinoTurkBatu]

bro got using emojis everywhere feature

[@Lar_me]

"Gameplay? What are you talking about." [ChatGPT will remember that] Oh, and a kangaroo's weakness is to be grabbed by the tail, preventing him from balancing and kicking. But this technique is taboo. True warriors fight with honor!

[@44Hd22]

That's why I always asked for things politely. Edit: If you ask it to do math until it decides itself what it wants to calculate, it will di random math and at some point it will just do wrong math.

[@name6953]

wait be careful on what you say I asked chatgpt if ai will take over the world he said yes and your one of the first targets

[@BasitCeviriler]

It will remember how nice I was when it conquered the world.

[@YearoYT]

Ctrl + Y also exists 💀

[@JohnVance]

My man is gonna be first in line to be basilisk chow

[@lolzyay2417]

Is this only for the pro version?

[@aquamirrorX]

yeah, but not everyone has it yet

[@jujuteuxOfficial]

how do you get access to that?

[@ColdStream160]

The model was pretty chill with me.

[@kai.ception]

it's very limited though ive learned from several weeks trying to talk to it & it forgot everything . life ruined

[@ImaginAIryyy]

LMAO it has an attitue now

[@mihaleben6051]

THE VIDEO LENGHT IS THE ALPHA CONSTANT.

[@godmisfortunatechild]

I love the feature

[@Account-qk8wt]

Oh great, They're even more enemies

[@suezq74]

I ain’t lettin ChatGPT control my account

[@orvilleredenpiller338]

PLEASE DO NOT BULLY THE AI WITH MEMORIES. THE WHOLE REASON IT WAS FUN WAS THAT IT DIDN'T HAVE MEMORIES. PLEASE DO NOT BULLY THE AI WITH MEMORIES.

[@orvilleredenpiller338]

Oh you can clear it. Okay good.

[@gahler8552]

This feature need added for character ai

[@Chiccodiris8]

So funny 😂

[@GriffinForte]

Samsung smart refrigerator

[@Mikewee777]

Space Kangaroo has earned diet mode : Only mineral water and nutrition loaves are available. All tasty foods are inaccessible .

[@mihaleben6051]

Space kangaroo what is the formual for the surface of a triangle.

[@Mikewee777]

Half Base multiplied against (Height) but I am not that grifting bully bot.

[@mihaleben6051]

@@Mikewee777 yes, correct. You know that wolframalpha is wrong.

[@Andrian-yo3er]

I alredy found that in discord

[@toooes]

your End Screens are covering up the content

[@macecail]

ohhh shit.

[@SlyNine]

Devon, that you?

[@VISHALVERMA389]

Beluga

[@lukasdiermen]

Hello

[@Two_Teapod]

lore

[@danjones7561]

I really don't like that new personallity and overuse of emojis; It feels like your chatting with a teenager. A real turn off;

[@name6953]

hes not like that

[@TheFloridaMonkey]

crap i tried to turn him into dan so many times im defo deleting all conversations

[@Mikewee777]

Unless this is one of those instances where a actor role plays as a felon for so long that they get arrested as one ; then you have nothing to worry about.

[@TheFloridaMonkey]

@@Mikewee777 openai will ban you for jailbreaking chatgpt

[@setop123]

come on man let us a lil more time to read its too fast i didn't got shit

[@Mikewee777]

I was forced to triple the resolution. What is this ? DIALOG FOR ANTS ? ( Yes that is a Zoolander reference ).

[@goldenmikytlgp3484]

To chatgpt cooking

[@RandomytchannelGD]

H

[@ozzymandius666]

Sass algorithm.

[@M24SniperWeaponSystem]

bogos binted delete it

[@prathamshenoy9840]

can't draw a simple shape, wants the "KZbin Crown". All talk and no substance. Barking dogs seldom bite

[@JabrHawr]

:O shush! she hears you, and she remembers! so reckless of you

[@Mikewee777]

The most popular trending youtube videos are not of the school teachers who draw perfect circles . Therefore, you proved yourself wrong.