ChatGPT Can SEE: Here’s How it Works!

In this video we explore the capabilities of ChatGPT with GPT-4V(ision)

Discover More From Me:

Resources From Today's Video:

#AINews #AITools #ChatGPT

[kiaranr]

I'm teaching myself robotics in my spare time and GPT vision has been huge. I often upload photos of PCBs, servos or any other component and ask questions about it. It has greatly accelerated my ability to understand the field and I never get stuck.

[DaxLLM]

I can imagine you could use chatgpt 4 v to help repair things around the house, say identify the components of a sink pipe. Once identified tell it that there is a leak at component A, how can I fix this? Etc.... Very cool Matt!

[HarryNicNicholas]

i got bard to catch someone doing an ebay scam, i saw a guitar i rather liked the look of, but i had no idea what model it was just "the make" so i asked bard about the serial number (i could see that) and it pointed out the make and model, and year, which didn't at all match the picture - so i showed it the picture and presto, it apologised for making a mistake and told me the ACTUAL make and model, i told ebay, the ad disappeared....

[JamesOKeefe-US]

This is so awesome. I took a picture of a lighted tree and asked ChatGPT to tell me how many lights were on it. There are 75 but it said 87...i didn't realize that there was a reflection in the background of the tree in a window and it counted the 12 lights reflected there as well. Scary good.

[JancobSweety-elkj]

Honestly, you seeing the same stuff I see and thinking of how to use it just makes me feel so happy.

[alexfitchcreates]

I've been using it to help me with my artwork by asking it for tips to improve upon it as I work through the process. I can ask it to interpret the artwork as well which is really fun for me lol. So many uses for it.. can't wait to watch this!

[MeMyselfAi-kgep]

Great content, your an inspiration to the Ai community! Keep leading the charge!

[hankbark]

Extremely insightful, Matt. Always love to see and hear from you!

[HarryNicNicholas]

8:00 i used to work for electronic arts in their purpose built building overlooking a lake, with a bar and cinema, in chertsey, south london, we were working on formula one 2000 and as mclaren were our neighbours we had kimi and couthard's f1 cars in the foyer - the stair case in reception was used in the film inception for the "never ending stairs" sequence of the movie.

[WrldOfVon]

Matt wtf is that thumbnail? 😂 just clicked on the video and I already know it's about to be a banger! Thank u for all you do within the ai community ❤

[ChrisSalemme]

Every morning I do puzzles in the newspaper. Today I was stumped on the word-search/crossword so decided to test out GPT 4V with a picture of the puzzle and asked it to find all the words for the specific clue. It found everything with no issues at all.

This is obviously an easy example, but cool to think I can start showing more complex puzzles and have Chat GPT solve them.

[philipashane]

Re: 4:51 “Take a deep breath”: I had heard that, in addition to this being kind of a funny thing to say in a prompt, has actually been shown to be quite successful in getting CGPT to do chain of thought reasoning for better results.

PS: love your work, Matt!

[ashkoreations]

Today I could not figure out what I clicked to enable a feature in Adobe illustrator and while I usually try to search for videos on how to fix it I took a picture of my screen and uploaded it to chatgpt and asked first what was on my screen and then asked how to disable it and it walked me through all of the proper steps. So totally agree that using it with your phone for what I generally use tik tok and YouTube for will come in handy😁

[Klokinator]

Always excited for major updates to AI tools.

[ronulbrich]

Inspired by your video, I took a picture of a German Document that included handwriting and asked it to translate it. It misspelled one handwritten word, but otherwise got everything else correct. This is a game changer for folks like me who are researching their ancestors where the records are in a foreign language and the script used is also from long ago. Thanks for your great video in opening our minds to ChtGPT Vision's tremendous capability; its a game changer.

[SoCalGuitarist]

I train stable diffusion LORAS and fine tune models. The captioning capabilities of GPT-V is soooo much better than everything else out there. Captioning went from the hardest part of my job to the easiest, almost overnight. What used to take me about an hour per 20 images of hand crafting keywords is now literally done in about 5 minutes with me really just dragging and dropping (and soon to be automated entirely once I have access to the GPT-V api). HUGE time savings.

[deevamassey]

Wow! This is amazing and a bit mind boggling. Thanks! I always enjoy your videos.

[CarletonTorpin]

A great use for the GPT 4 Vision was going through my phone's photos and looking for images I'd taken of notebook-pages with writing on it and dry-erase boards full of text.

[countvoncount]

I took a picture of an interesting location and gave this prompt:
Let's play a game. You act as a text adventure and base the current scene on as much information you can gather from the image. You then give me prompts as in an oldschool text adventure, e.g. when i say something like "look around" you describe the scene to me. If I say "help", you give me hints at what I could say.

[Bartskol]

I think the most useful thing would be diy projects and simple repair work around a house where you can photograph a scene and tools and explain a task and chat can break it into a tasks. Exacly as Openai presented. Im waiting for teacher ai with personalized lessons and some sort of direction underneath the chat.