OpenAI’s ChatGPT with superior performance has had the most media attention of any AI chatbot on earth. and it has a better UI, better model, multimodal, and usability This was well deserved for some time, as the most powerful example of its kind, but this is a hotly contested pedigree. We pit ChatGPT vs Apple Ferret AI models head to head, and the results are surprising.
ChatGPT vs Apple Ferret 7B
We’ve entered an era of marketing in the AI chatbot space in which LLMs (that handle text input and text output) are largely understood by the enthusiast crowd as the source of an AI’s power. We’re no longer cutting-edge early adopters and instead are firmly in the mass adoption and education stage of the Gartner Hype Cycle. As a result, these large language models are being used to distinguish one chatbot from another, often flaunted like a badge of honor. GPT-4 is the achievement in OpenAI’s effort in scaling up deep learning and has had far more media exposure than any other AI model and is the only one that you could even hope that someone outside of the tech industry might know by name. Multimodal Large Language Model (MLLM) Ferret is in a very good position can understand the spatial reference of any shape, coordinates size, or granularity within a picture.
The danger here is that GPT-4, and the Generative Pre-trained Transformer line now trademarked by OpenAI, could become ‘stylish’. There’s an unavoidable shortcoming of the human psyche that we assign social value and prestige to even the most objective of tools. Its reputation precedes it, and that reputation will be maintained past the point of function, as new and more powerful competitors rise to the challenge. The most popular tool is not always the most powerful, and thanks to benchmark results we can see exactly how the two match up. Google Bard, Samsung Galaxy AI, Grok AI, Meta Chatbots, Microsoft Copilot, and more are Apple Ferret’s direct competitors in the AI race.
Apple’s new base model Ferret 7B and 13B are trained on massive custom hardware- 8 Nvidia A100 GPUS with 80GB memory and have impressive multimodal capabilities. Apple’s GitHub repository offers open-source access to its two different AI models. Both models are trained using the dataset (LLAVA CC BY NC 4.0 allowing only non-commercial use) which is not used outside of research purposes. Apple’s hybrid region representation with features like a spatial-aware visual sampler for AI Apple researchers has introduced Ferret, an innovative multimodal large language model (MLLM) specialized for referring and grounding in images. With the Ferret model, Apple’s dedication to advancing Artificial Intelligence and offering innovative solutions with practical applications in multimodal conversational interfaces. Apple has to build better real consumer products and compelling consumer product that can only function with AI where Apple can edge out every single competitor.
Does Apple have an AI chatbot?
The most powerful LLM on earth is of no use to the average person without an accompanying AI chatbot. This is the user interface (UI) through which you can use the language model, and the elephant in the room here is that Apple’s LLM (Multimodal Large Language Model–MLLM) does not yet have an AI chatbot. It’s a no-brainer that CEO Tim Cook will have one in the works, though.
| Company | CEO | AI Chatbot | LLM | API | Open-source |
|---|---|---|---|---|---|
| xAI | Elon Musk | Grok | Grok-1 | No | No |
| OpenAI | Sam Altman | ChatGPT | GPT-3.5, GPT-4, GPT-4V, or GPT-4 Turbo | Yes | No |
| Sundar Pichai | Bard | PaLM 2 | Yes | No | |
| Microsoft | Satya Nadella | Copilot (formerly Bing Chat) | GPT-4 | No | No |
| Meta | Mark Zuckerberg | Meta AI | LLaMA 2 | No | Yes |
| Anthropic | Dario Amodei | Claude | Claude-2 | Yes | No |
| Amazon | Andy Jassy | Olympus (rumored) | Olympus (rumored) | No | No |
| Apple | Tim Cook | Ferret (7B or 13B) | No | Yes |
Apple Ferret models are expected to perform better than GPT-4 when run natively on mobile devices, as they were intended to be. The compact size (by modern standards) of 7 billion parameters is suitable for a mobile platform like iOS (iPhone) and Android phones, whereas GPT-4 has in excess of 1 trillion parameters, making it substantially more powerful (for e.g- Depth Fusion powered by ML models in latest iOS camera app takes more crisper photos), but limited to server implementation and only accessible with an internet connection.
Essential AI Tools
Final thoughts
ChatGPT has had a great run, but the competition today is much fiercer than when it launched back in November 2022. Its place in the public zeitgeist doesn’t necessarily reflect its ranking anymore. In all fairness, there are several other factors to consider, including plugin support, which provides such a wide variety of functionality and software integration that it could still be called the best AI chatbot, even if it were the 2nd or 3rd most powerful. In summary, we should avoid assigning ourselves a ‘fan’ of any one tool, and instead be prepared to evolve along with the technology itself.
