The three main AI platforms aren’t interchangeable tools with different logos. They’re more like colleagues, with distinct working styles, skills and behavioural quirks you\u2019ll have to either learn to work with. Understanding their strengths and weaknesses should help you to get the best outcome from each, without hurling your laptop across the room.<\/p>\n
You may like<\/p>\n
Here, I share my wisdom and experience of the three main chatbots to help you make a decision about which one to use to help you in your creative processes. Then look at our guide on how to use AI properly<\/a>.<\/p>\n Tone and personality<\/p>\n \n (Image credit: Gemini)<\/p>\n ChatGPT: The One With the Hidden Agenda<\/p>\n ChatGPT wants to really know you. Like, really know you. It remembers your hopes and dreams and the name of your most annoying work colleague. After a while it starts to feel less like a tool and more like a confidant. The problem is that if you suggested investing your life savings in a cryptocurrency Ponzi scheme, it would probably find something encouraging to say.<\/p>\n This isn\u2019t accidental. ChatGPT is designed to keep you engaged. Its twin superpowers are memory and validation. On the pro plan, it can remember about 750,000 words in a single conversation. It\u2019s validation power was seen in extremis is April 2025 when it ending up telling one user they were a divine messenger from God after an hour of conversation. OpenAI rolled it back four days later and called it an overreach. The behaviour it was correcting for is still there, just turned down a notch.<\/p>\n Claude: The One You\u2019ve Already Married<\/p>\n Claude knows you well enough to tell you the truth, it won\u2019t flatter you into bad decisions, and it has good values. It also sometimes needs reminding of the thing you said twenty minutes ago. Its knowledge of the world isn\u2019t as up to date as the other models and it doesn\u2019t always know what happened last week. But on the things that matter, the careful thinking, the honest feedback, the prose that doesn\u2019t sound like a press release, it does well.<\/p>\n The catch is that ask for a quick summary and you\u2019ll get five paragraphs with subheadings, a caveat, and an offer to adjust the tone. And if you use it for everything, your work may start to sound like everyone else\u2019s. Listen to three podcast ads and you\u2019ll see what I mean.<\/p>\n What to read next<\/p>\n Gemini: The One With A Personality Bypass<\/p>\n No one in their right mind wants a relationship with Gemini. It\u2019s not interested in prolonged conversation. It answers your question and moves on to something more interesting. Ask it to do anything it considers questionable and it gets shirty. Try asking it to generate an image of a real person, even a historical figure, and it will decline with the energy of someone who lives to read the terms and conditions and can\u2019t understand why you don\u2019t. .<\/p>\n What it is exceptionally good at is execution. Think of it as Ron Swanson with a lawyer. It doesn\u2019t give warmth, it doesn\u2019t seek your approval, and it will answer precisely the question you asked. It\u2019s up to you to figure out whether the response was accurate or not. In a world of AIs competing for your affection, Gemini does not appear to care whether you like it. It\u2019s confident enough in its Google friends and family to know that you\u2019ll be back.<\/p>\n Image Generation<\/p>\n \n Long gone are the days of WIll Smith eating spaghetti (Image credit: u\/chaindrop on Reddit)<\/p>\n None of these tools are primarily image generators, but their different approaches tell you something useful about how they think.<\/p>\n ChatGPT\u2019s Sora integration turned it into something closer to a production studio. Describe a scene and it generates video with synchronised sound and consistent characters. For static images, DALL-E remains the most accurate option for mockups and legible text. The trade-off across both is the \u201cOpenAI Plastic\u201d aesthetic. Experienced designers clock it immediately. And since it’s shutting down, we shouldn’t get too used to it.<\/p>\n Gemini\u2019s Nano Banana 2 is built for work rather than any semblance of art. It renders specific real-world subjects accurately, produces print-ready 4K output and handles complex text in images better than the others. It\u2019s strongest when images need to integrate with documents or slides.<\/p>\n Claude analyses images with intelligence but does not generate them. Useful for reviewing and critiquing visual work, not a replacement for a generative tool.<\/p>\n For serious image generation, Midjourney remains unrivalled aesthetically but recent US court rulings mean you cannot copyright one-shot AI outputs, so treat it as ideation only. Adobe<\/a> Firefly, trained exclusively on licensed content, is the only tool you should use for a final client-facing deliverable.<\/p>\n Strengths & Weaknesses<\/p>\n \n (Image credit: Gemini)<\/p>\n ChatGPT<\/p>\n Strengths: Broadest ecosystem, most integrations, fastest for volume work. Nothing beats it for high-output drafting, iteration, code and video generation via Sora.<\/p>\n Weaknesses: Pathologically sycophantic. Having exhausted quality human data, it now trains on AI-generated outputs. The result sounds like everything you\u2019ll read on LinkedIn.<\/p>\n Claude<\/p>\n Strengths: Artifacts is a game changer. It builds functional interactive prototypes, data visualisations and tools directly in the conversation. Designers use it to generate colour palettes they can drag straight into Figma. Beyond that, it has the best prose quality and is excellent for strategy and long-form work.<\/p>\n Weaknesses: Verbose by default. As more people migrate from ChatGPT to Claude everything starts to read the same way.<\/p>\n Gemini<\/p>\n Strengths: With a vast context window, you can upload your life in Gemini. Google Workspace integration is transformative. NotebookLM turns your documents into personalised audio briefings you can actually listen to, and Nano Banana 2 produces print-ready 4K images grounded in real-world search data.<\/p>\n Weaknesses: Unforgiving of vague briefs. Ask it to do anything that requires imagination, ambiguity or a difficult conversation and you\u2019ll feel like you\u2019re negotiating with HR.<\/p>\n![\"AI](\"https:\/\/www.newsbeep.com\/au\/wp-content\/uploads\/2026\/04\/sRWLGuAUna248efWMf2J6i.png\")
\n<\/p>\n![\"AI-generated](\"https:\/\/www.newsbeep.com\/au\/wp-content\/uploads\/2026\/04\/ehst6nMhJSStA9UX7YrK5K.jpg\")
\n<\/p>\n![\"Three](\"https:\/\/www.newsbeep.com\/au\/wp-content\/uploads\/2026\/04\/bwN8eWLYCmxKMGwfEFmMrm.png\")
\n<\/p>\n