The Nuances of GPT-3 as a Personal Assistant: Potential and Pitfalls

‍

You have a use-case in mind for GPT-3 personal assistant?

‍

Hey guys, if you're reading this it means one thing:

‍

• You understood the role AI can play in your workflows.
• You want to build a GPT-3 personal assistant for a specific use case but don't know where to start.

‍

At Argil we're currently building the next version of our product in which:

‍

• You can chat with different expert personal assistants.
• Give them skills such as speak with a document and interact with a webpage.
• Train the AI on your pictures if you want to generate specific images.
• Get multimodal outputs based on text entries. (Images + text)

‍

If this resonates with you and you wish to learn more about it, please send me an email explaining what Gpt-3 personal assistant do you need and we'll see how we can build it for you :).

‍

Contact me at the following email: othmane.naimkhadri@argil.ai

‍

GPT-3 challenges the economic state of society

‍

In just a year, the economic state of our societies has been completely challenged.

‍

The launch of GPT-3 will be remembered in history as a massive shift in the interaction between:

‍

• Work.
• Humans.
• Technology.

‍

OpenAI has garnered substantial attention for the capabilities of LLMS.

‍

It showed us that the realm of applicability is wider than we expect, and that it can definitely work in hand with us in our daily tasks.

‍

What is sure is that we can say that:

‍

GPT-3 personal assistant vision was born with OpenAI.

‍

Why People Use GPT-3 as a Personal Assistant:

‍

GPT-3 gave people a way to interact with a massive corpus of data in a quasi-instant way.

‍

The shift is massive and completely reshuffled the card of creation:

‍

• Ask questions about history to GPT.
• Use it as a guide in your brainstorming sessions.
• Use it to derive insights/summaries from data inputs (Excels, Documents)

‍

This vision of gpt-3 personal assistant is based on a massive component:

‍

The ability from people to prompt in a good way.

‍

If we go one step further, GPT-3 solves 2 major Challenges:

‍

• Scalability.
• Time Allocation.

‍

GPT doesn’t stress, GPT doesn’t rest, GPT doesn't ask for a rise.

You can use it whenever you want and give it massive workloads, it will deliver.

‍

That's the reasons behind the GPT-3 Personal Assistant vision.

You can now focus on the strategic tasks of your company and delegate many tasks in your daily workflows.

‍

How does GPT-3 personal assistant gives outputs?

‍

The mechanism behind gpt-3 capabilities is based on 2 components:

‍

• Prompt Analysis.
• Response Generation.

‍

Most of the time and for not too complex tasks, GPT-3 will comprehend your intent and give you an output you’ll be satisfied with.

‍

As it’s trained on a massive dataset, GPT-3 can help you in all type of tasks:

‍

• Draft emails for you.
• Write your LinkedIn posts.
• Summaries articles or emails.

‍

All the day-to-day micro tasks in which you believe you’re waisting time, GPT-3 can become your Personal Assistant in.

‍

But GPT-3 has it’s flaws and does not always fit the corporate agenda and requirement.

‍

Big companies required tools with more granularity and transparency. Which is not the case of GPT-3. We do not have access to the dataset on which it has been trained.

‍

On top of that GPT-3 is not customisable to the end-users of companies, making it a hard choice for a personalized experience.

‍

Without even mentioning the regulatory framework that will require specific verifications and conformities in regard to the data privacy.

‍

Current Use-cases of GPT-3 Personal Assistant

‍

If you didn’t use GPT-3 as a personal assistant yet here’s a list of use-case you should definitely try it for:

‍

Idea Generation:

• Brainstorming made simple, aiding in curating content across platforms.

‍

Drafting:

• Crafting communications, be it internal emails or outward-facing customer support.

‍

Optimization:

• Improve content by rephrasing or assigning efficiency scores.

‍

Interview Prep:

• Be it for hiring or making sales, GPT-3 tailors the perfect set of questions and answers.

‍

Quick Summaries:

• Information overload? GPT-3 distills articles, presenting just what you need.

‍

Translation:

• Transcend language barriers effortlessly.

‍

While GPT-3 ticks many boxes in the Personal Assistant role, when personal touch and scalability come into play, it fails to provide a convenient experience.

‍

What is needed to improve the GPT-3 Personal Assistant vision

‍

Here’s what would be needed:

‍

• Training over datasets.
• Contextual understanding.
• Easier user interface and user experience.
• Open to collaboration between the same collaboration.
• Multimodal approach (Interact with webpage, Ask questions to document, generate images)

‍

For that to happen OpenAI would have to become a product company which I believe is not really their intention. One can understand that just from the strategic move they took with the GPT-plugins.

‍

If they wanted to be the one building on the GPT-3 personal assistant vision, they wouldn’t:

‍

• Open an access to their API.
• Let people build their plugins.
• Propose the plugins directly in chatGPT+.

‍

Customization and collaboration on top of chatGPT won’t come from OpenAI but from companies building on top of the foundational model.

‍

Juste like us at Argil:

‍

The Evolution of GPT-3 Personal Assistant: Argil.

‍

At Argil we see the next step in the personal assistant vision as being the Anthropomorphising of ChatGPT:

‍

• Each of your discussion is on a project.
• Each of your discussions call to a specific expert.
• Each of your experts will have a specific set of skills.

‍

On top of that we deeply believe that the multi-modal approach with skills is a must:

‍

• Generate Images.
• Ask questions to web pages.
• Ask questions to your excels.
• Ask questions to your document.

‍

At Argil we’re building a GPT-3 personal assistant on steroids.

‍

If you want to be part of this adventure, join us: here

‍