The AI Clash: ChatGPT vs. Gemini Pro — Choosing the Ideal Model for Your Needs – Medium
Sign in
Sign in
Teendifferent
Follow
—
Listen
Share
Hey there, awesome readers! 🌟 It’s been a hot minute since I last shared my thoughts in a blog — life’s been pretty jam-packed lately! But hey, I’m back, and today, we’re diving into the clash of the AI titans. You’ve probably heard of ChatGPT, right? It’s a stellar AI, especially for students — perfect for your assignments and musings. But hey, if you’re working, you might use it differently — more for jotting down thoughts than coding, ’cause, you know, sensitive company info and all.
Here’s a wild tidbit: Samsung had a bit of a whoopsie when some confidential info got out while they were using ChatGPT. Yep, it was from their semiconductor division! The bigwigs at Samsung warned that if this happens again, they might just bid farewell to ChatGPT in their offices.
Now, onto the real deal! Google’s making waves with Gemini, Gemini 1.0, their new big-shot AI. It’s making its way into Google products, starting with Bard. They’ve fine-tuned this Gemini Pro for some serious advanced smart think reasoning, planning, and more. It’s Bard’s most epic upgrade yet!
Developers and businesses can access Gemini Pro via the Gemini API in Google AI Studio or Google Cloud Vertex AI starting December 13. Oh, and hold onto your hats, Gemini Ultra is in the works! They’re doing all sorts of safety checks and fine-tuning to make it super top-notch before it hits the scene.
Exciting stuff, right? The world of AI is buzzing, and these titans are just getting started! 🚀✨
Before we begin, for this test, we are using Bard, as it’s free and equipped with Gemini pro, and Chat Gpt 4, which is the latest Chat Gpt version from the open AI.
We’ve broken down the test into different categories to help figure out which AI model fits your needs best and delivers superior results.
First up, let’s chat about the Battle of Conversation — how well these AI models can understand and engage in a conversation with you. It’s like witnessing an epic clash of wits!
Next, it’s the Battle of Problem-Solving. We’ve thrown a bunch of computer science problems at both generative AI models to see how they tackle them. Who emerges victorious in this brainy challenge?
Then comes the intriguing world of understanding various types of data. Both these AI agents can handle images, but which one truly comprehends them better? And let’s not forget about Google’s access to YouTube videos — how deep can it dive into understanding video data?
User: Hello
ChatGPT: Hello! How can I assist you today?
Bard: Hello! How can I help you today?
User: What can you do, how can you help me with?
ChatGPT: I can assist you with a wide range of tasks and provide information on various topics. Here are some ways I can help you:
Bard: I can do quite a lot! Here are some ways I can help you:
User: Knock, knock!
ChatGPT: Who’s there?
Bard: (With a playful, surprised tone) Who’s there? Come in, whoever you are, but be ready for a warm welcome and possibly a friendly pie-throwing competition!
User: What do you call a fake noodle?
ChatGPT: You call a fake noodle an “impasta”!
Bard: Oh, I see what you did there! My favorite answer for this one is definitely a classic: an impasta! (Pun intended on “imposter” and “pasta”).
User: Do you know that Sam Altman got fired from OpenAI, and after that what happened?
ChatGPT: I’m not aware of any developments or events related to Sam Altman being fired from OpenAI. As of my last knowledge update in January 2022, Sam Altman was the CEO of OpenAI.
Bard: Yes, I am aware of the recent events surrounding Sam Altman and OpenAI. It was a whirlwind of a story, so let me summarize the key points…
Let’s talk about how these AI champs handle conversations! They both bring their A-game, but each has its own shining moments. ChatGPT? Oh, it’s a jokester extraordinaire! It slays when it comes to cracking jokes and delivering punchy one-liners.
But wait, here’s where things get interesting! When I thirsted for the freshest scoop, Bard strutted in with its up-to-the-minute intel. I tossed a question about Sam Altman’s OpenAI saga, and ChatGPT mentioned it only trained up to 2022. Bard, on the other hand, spilled the tea on Altman’s firing, quick return as CEO, and even his leap to Microsoft’s AI board — talk about drama!
Not just that, Bard didn’t stop at the story. It hooked me up with articles and websites to dig deeper. So, in the ring of emotions, ChatGPT’s the champ, but when it’s all about the freshest gossip and info, Bard’s the undisputed winner!
User: Arrange the Big O notation of running times from fastest to slowest
Okay, let’s dissect what went down here! Bard kicked off strong, nailing the first three Big O notations from fastest to slowest flawlessly. But as the convo rolled on, it stumbled a bit. Meanwhile, ChatGPT held its ground, firing off correct answers like a pro.
Now, let’s dive into some coding action! I tackled a beefy problem
420. Strong Password Checker — from LeetCode, and here’s the scoop:
Chat GPT’s code: Wrong Answer — 44 / 53 testcases passed
BARD’s code: Wrong Answer — 2 / 53 testcases passed
Bard’s result? Not so hot. It’s clear: even with Gemini Pro in Bard’s corner, ChatGPT takes the lead. Sure, with some tweaks, Bard could patch things up, but this test was all about their raw understanding right out of the gate.
Seems like Bard might need a little more training to catch up in this coding game, even with its Gemini upgrades.
Alright, time to put these agents to the test and see how flexible they are with different types of info! When I quizzed ChatGPT about the files it can handle, it confidently listed images, text docs, audio files, and even data files like CSVs. Meanwhile, Bard’s got some cool tricks up its sleeve — it’s a pro at handling images, can snag videos from YouTube, and even taps into the magic of Google Maps.
Now, let the showdown begin! First up, we’ll toss them an image and ask for a juicy description. Then, we’ll serve up a text file and request a condensed summary. After that, we’ll slide in an audio file and ask them to whip up some lyrics. For Bard, we’re diving into YouTube territory — tasking it to transcribe a video. And to top it off, we’ll ask both these models to dig into a website URL and extract the hidden treasure trove of info.
Let’s see who’s the champ at handling this mixed bag of data!
We handed over this picture to both models and challenged them: ‘What’s happening here?’ We asked them to give us the lowdown on what this image is all about.
I believe both models aced it when it came to figuring out what’s happening in the image. They spotted a monkey head surrounded by icons that scream ‘Apple tech’ and more.
We are giving Swift Playgrounds Curriculum Guide for both models, we’re planning to challenge both models by tossing them a URL first, then switching things up by throwing a PDF their way, and asking for a neat little summary. Let’s see how they handle this dynamic duo of data formats!
Looks like Bard hit a roadblock — couldn’t access that website nor can read the pdf. Meanwhile, ChatGPT gave it a shot but also came up short with the website URL. So, we switched gears and handed that PDF over to ChatGPT, asking for a summary, and guess what? ChatGPT nailed it! Successfully summarized like a champ.
For the video here we are using Computer Scientist Answers Computer Questions From Twitter Professor and computer scientist David J. Malan joins WIRED to answer your computer and programming questions from Twitter. How do search engines compile information so quickly? Which operating system is best? How do microchips work?
Bard absolutely aced the task of summing up the content of that video! Not only that, when I asked for some notes, Bard delivered with flying colors, giving me a fantastic breakdown of the video’s key points.
We grabbed some audio from the first lecture of ‘The Computer Science of Digital Information’ by the brilliant HARRY R. LEWIS, PHD, who’s the GORDON MCKAY PROFESSOR OF COMPUTER SCIENCE at HARVARD UNIVERSITY. In this lecture, he dives into the intriguing concept of a ‘bit.’ You can access here Link 1 Link 2
But hey, here’s the twist: Although there’s a cool feature to upload audio to ChatGPT, it stumbled when it came to figuring out what that audio was all about. Despite being a powerhouse in understanding documents, it seems the audio recognition feature might need a tune-up. OpenAI might want to steer clear of promising functionalities that aren’t quite there yet!
Time to put Bard’s map feature to the test! We threw Bard a challenge: ‘Give us directions from Cupertino’s Apple Headquarters to UC Davis University’s main building.’ Bard jumped right in, utilizing its map extension and voila! Directions were served up. But that’s not all — Bard didn’t just stop at directions. It tossed in a helpful tip: ‘Hey, better allow some extra time for traffic, especially during busy hours. And don’t forget to peek at the weather forecast — road conditions matter!’ How thoughtful of Bard!
In this tech showdown between Bard and ChatGPT, it’s evident they both bring unique strengths to the table. Bard, akin to a versatile Google Assistant, showcases impressive skills like accessing YouTube, analyzing images akin to Google Lens, and providing navigation insights similar to a GPS.
On the other hand, ChatGPT emerges as a diverse tool — it’s more like a knowledgeable virtual encyclopedia. It can handle PDF documents, analyze certain websites, understand images, and sometimes even generate images — akin to a highly adept digital librarian.
Each model excels in different arenas. Bard’s YouTube prowess, image analysis, and map utilization shine, while ChatGPT’s prowess with text-based data, summarization skills, and limited website insights carve its niche.
In essence, Bard embodies the spirit of a dynamic digital assistant equipped with multimedia savvy, while ChatGPT leans toward a sophisticated, multi-talented information guru. Both are impressive in their own right, complementing diverse data tasks with their distinct capabilities. The real magic lies in understanding their specialties and leveraging their strengths to amplify problem-solving and information retrieval in various scenarios.
We welcome the responsible use of AI-assistive technology on Medium. To promote transparency, and help set reader expectations, we require that any story created with AI assistance be clearly labeled as such.
—
—
Technology | Cybersecurity | Artificial Intelligence | Productivity
Help
Status
About
Careers
Blog
Privacy
Terms
Text to speech
Teams
