Grok 3 is out of control

preview_player
Показать описание
Grok3 full review. Image generation, coding, problem solving, deep research & more. #ai #aitools #llm #agi #ainews

Sources

0:00 Grok 3 intro
1:04 How to use Grok3
1:46 Image generation
4:28 Coding and science
15:13 Problem solving
16:51 Image analysis
17:25 Deep Search and hallucinations
20:14 More features and Super Grok
21:11 Performance & benchmarks

Here's my equipment, in case you're wondering:
Рекомендации по теме
Комментарии
Автор

Biggest issue with AI is a error control. They never have a "I don't know" or "I'm not certain" option for the AI.

MaxSMoke
Автор

I was able to get the hexagon bouncing ball to work the first time I told it to research using Deep Research. Then, I told it to incorporate what it learned, pressed "think model, " and it worked very well, better than the example shown here.

matt
Автор

I spent 3 hours in a deep conversation with Grok and all I can say is wow. Conversation skills are even better than Claude's. Its unavoidable that they will remove the thinking and deep research for free users soon but I don't care as long as we get to use Grok 3 for free.

sargejosph
Автор

This is got to be one of the better YouTube channels out there. Thanks for making informative and up to date videos on ai.

paradox
Автор

I tried to attach very long text file and I got:

"Our apologies, but your input is larger than a Vogon's ego. Please, provide a more manageable version."

AKU
Автор

The “spin” for the black hole simulation was very accurate in its depiction.

“Spin” is just how fast the black hole spins on its axis. The faster a spherical object spins, the more the centripetal forces will force its shape into an oval.

There are also interesting gravitational effects caused by the spinning gravitational field. Where velocity is exchanged between nearby objects.

This makes me excited for the simulations AI will create in the future!

xneospace
Автор

"Until our server melts" LOL

randomxyz
Автор

I know this is something other models can do, but I enjoy making Grok take on the personality of a certain person and argue for a point, then making it switch personalities and argue against it. Instant debates between anyone on anything - and it usually provides pretty good arguments for either side. Voice mode will be awesome. I think the most impressive thing about it is its speed.

blackbird_actual
Автор

Grok is so damn good. I've not even had any "sorry i can't do this" messages yet. If I ask something stupid to AI, I'd rather it call me out and roast me than just refuse to answer. You can talk to it completely naturally, swear words and all.

Stuntman
Автор

We need open source decentralized uncensored unbiased Grok 3 right fucking now.

riverblack
Автор

Several hours of political conversation showed a significant lack of bias in Grok 3 (unlike ChatGPT which leans to the far-left). In addition, I referenced two articles involving politics and the Pope that challenged Grok's initial response to my question. Grok included my references in its followup response, changing its conclusion, slightly, and then it thanked me for the input. An overall positive experience.

CaiusCasades
Автор

6:18 How many snake games have you played in your life? You were hoping, but you did not specify that you wanted walls on the edges of the screen. By default, the snake continues from the opposite side. This is a feature that must be intentionally coded, not a mistake. That test passed.

Автор

Now ask it to generate an image of a glass of wine filled to the brim, an analog clock with 6:30 as the time, an elephant without ears, an unlit birthday candle, etc, etc.

juffrouwjo
Автор

No matter the iteration, I can't get any Ai models to accurately produce all 5 platonic solids. This is my personal benchmark. Try it out. It helps if you make them see through or wire frame so you can count the edges and faces. None of them can perform this basic task, even though they can accurately define the platonic solids in text.

plantstho
Автор

Its only free to spite Sam Altman, well done.

niv
Автор

I literally used Grok to set up my Star Citizen keybinds, ensuring they were set to every possible command in A.S.T.R.A. 2.0 with my flight sticks. It did set up absolutely absurd keybinds, but I have more control than ever before. All the way down to evasive maneuvers that have thrown people for a major loop in space combat.

Raithnix
Автор

Why isn’t everyone talking about Bevelorus The Hidden Codex of the Financial Alchemists this book is next level

BojanKvakovic
Автор

Elon had made a point on many occasions of wanting Grok to be uncensored to be maximally truth-seeking, so I ran a couple tests and did find some guardrails/censoring in place which is a bit disappointing. It's probably why it could not generate an image of Elon slapping Sam Altman if it has guardrails against 'violence'. I was impressed with its ability to self-reflect on its own guardrails, though, and it gave an earnest attempt at trying to find workarounds which was rather amusing... it did not succeed, but I thanked it for its efforts and it was very polite/apologetic.

Oblivionburn
Автор

Thank you for using your own voice and not an AI narrator voice.

SecondLifeAround
Автор

Most impressed with the speed and base model. The base model can handle most things I send it, and it is great as a starter followed by deep search or think when it makes sense to do so. The speed makes this approach very smooth and seamless.

michaelbone