The pioneers of A.I. have been many, including John McCarthy, Douglas C. Engelbart, Alan Turing, and Ada Lovelace, among countless others.Great ideas are built on top of previous great ideas - in the story of augmenting intelligence - it's no different.
/Augmenting Memory/
Humans developed language by using (small mouth noises) and attaching meaning to them. These symbols were held in collective memory and passed down generation to generation. Language, I would argue, is our greatest Tech. It plays a crucial role in all human processes. Through language, we express our thoughts, share knowledge, and document wisdom. Symbols can take various forms, including spoken words, written text, and internal mental representations.
The development of written language was a significant milestone in augmenting memory. Stories were no longer held solely in the mind(s) of our ancestors. It allowed knowledge to be recorded, preserved, and transmitted over longer spanses of time. We know little about early writing systems though various forms seemed to have emerged in different civilizations independently. Preserved texts have been critical in helping societies build upon the insights of the past.
/Prompt Engineering/
It's a buzzword term referring to interacting with A.I. models to extract specific outputs. Language models generate responses based on the input they receive, which includes a prompt or (chained prompts). This is not limited to text or code generation - but includes visual models such as Midjourney.
All prompting requires understanding the capabilities and limitations of the A.I. model(s) and creative thinking to design prompts that maximize the usefulness of the generated content.
However, considering current LLMs are I/O machines, everything generated must undergo human validation. It still comes down to how effective is the human agent and what their expertise in a given domain.
/Brevity/
In the realm of communication and information exchange, brevity is paramount. Less is more. Simpler the better. Conveying ideas with this heuristic enables smoother communication with A.I. systems. Even basic awareness of synatx using (), {}, and [] help to signal important information. Brevity helps to hold attention. Transformer models rely on an attention mechanism to streamline information, and deliver messages clearly and succinctly.
/Local LLMs/
Local LLMs are an OSS (open source software) approach to bring the power of LLMs into a local dev environment. These are lightweight versions of LLMs running 4-bit, 5-bit, or 8-bit integer quantization support. Allowing developers to play with, utilize, and fine-tune their capabilities.
Local LLMs allow for much faster inference speeds, and can be run at a fraction of the cost of larger hosted models. But to be fair - they are possible because 100s of billions were invested in scrapping, training, and creating the large foundational models, Chat GPT, Llamda, Bart, Jurassic, etc. The ability to leverage the power of language models for specific tasks without the need for large-scale infrastructure or extensive computational resources.
This offers flexibility, scalability, and efficiency in deploying language models for various needs Anton is a developer on the cutting edge of local LLM development.
/API Relient/
Why not just build an application using the APIs (application programming interfaces) of foundational A.I. models? API reliance is a fine line to walk. In some instances, using an API is the only option, i.e., (Experian Data) for consumer credit. However, dependence on API access for core features, functionalities, or services is a risky business model.
In the context of A.I. and language models, API reliance refers to using APIs provided by language model platforms, such as OpenAI's GPT, and building a wrapper around a specific model. Jasper did this and raised 120 million, but my predication is they will go bankrupt in < 2 years.
/Conclusion/
There is a non-zero chance that A.I. will be the end of humanity. It could also be our greatest triumph. The future is uncertain, and the ever-evolving digital landscape.
Labor has been a limiting factor since time immemorial. If A.I. handles some of the cognitive load - perhaps it can drive human prosperity. A hammer can be a weapon or just a hammer. My only hope is that some wisdom and luck will steady our shaky hands.