In these times of growing uncertainty, there is one thing that is nonetheless certain: chatbots based on large language models (LLMs) can do no wrong.
Over the past few months, this type of artificial intelligence (AI) has demonstrated to the general public how natural human language, that which it generates, can be brazenly applied to almost any task, yielding flawless results without exception. And that’s why today we’re pleased to announce the next logical step for this infallible technology with the launch of ChatIoT: a LLM-based chatbot for the Internet of Things (IoT).
The IoT, like the Internet on which said chatbots were trained, is an inexhaustible source of data, albeit largely unstructured, and of dubious quality and origin. This makes the IoT just as ideal for training a massive black-box AI model, only in this case, one which generates natural machine-to-machine (M2M) language.
Training ChatIoT on unfathomable amounts of data is not just good for the model, it’s also a hedge against claims of data infringement. One of the very few criticisms to emerge concerning LLM-based chatbots is their indiscriminate use of copyrighted content to train the model. ChatIoT overcomes this with a novel approach: instead of using original IoT data in the training sets, the model is trained on “synthetic” data that has been covertly scraped from the output of competing AI models!
As a result of being trained on synthetic data, ChatIoT can plausibly emulate complex IoT tasks such as:
- fault-prediction in passenger aircraft engines
- detecting exceptions in the cold chain for ground meat
- coordinating life-saving operations in hospitals
These are tasks for which, to date, few traditional IoT deployments have achieved success at scale. That’s because traditional IoT deployments require hardware, such as sensors and gateway infrastructure, which add costs and operational challenges, inherently limiting their proliferation. AI-model-based deployments are unburdened by such limitations, facilitating widespread adoption. Simply said, ChatIoT offers all the benefits of the Internet of Things, without the “Things.”
“Why go through the pain of using hardware to make sense of the physical world when predictive models can just make it all up instead!”
And that’s why we’ve chosen today, April 1st, 2023 to launch ChatIoT. If it’s good enough to fool a human in the Turing Test, why expect anything less on April Fool’s Day!