💫 Co-bots: how to think of the synergy between Humans and Machines

From rivals to allies, how software can help and how companies should think of implementation in the workplace

👋 Hello, and welcome to this new edition!

Today, we will cover the rise of ‘Co-bots’: how AI will make software designed to work alongside humans, rather than replace them.

You’ll find out:

• ➿ Why we need ‘humans’ in the loop

• 💫 What are ‘co-bot’s and how their role is to inspire, mentor, collaborate

• 🌊 Solving for abundance of information for learning

• 📏 How the key will be personalization and regulation for companies implementing this technology

Enjoy!

➿ Humans in the Loop

Everyone's scared of AI taking their job. But if you don't consider extreme cases, it will be a question of taking away tasks, not jobs.

According to research by OpenAI, 80% of US workforce will have 10% of their tasks displaced. Only 20% of workers will have 50% of tasks displaced.

While the panic and the novelty is spreading, companies and employees who will benefit the most of this technology will be those who already identify what that 10% looks like exactly, and figure out which tools to use, and how to get better results - now. That itself, is not an overnight feat. You don't give a job to the AI and forget about it completely (at least not yet).

There will always be the need for a 'human in the loop':

Responsibility - Regardless of automation, we've already moved into a world that will focus on results and outputs, not time. This means that you will be always responsible, as a human, for the output.

Personalization - We're at the beginning of a new era, and while it's unregulated (both from a policy but also from an 'ethical' point of view) we will be experiencing a flood of automatically generated content. Where's the added value going to be? In the human touch.

Hallucination and Bias - Fake information, biases. Yes, they're being worked on, but may not be going away. Are you going to blame a machine?

💫 Co-bots: Mentors, Inspirers, Collaborate

By turning this perspective, we should see any form of AI as a part of this ‘loop’ or system, as a tool. That’s why we should call them ‘co-bots’.

Co-bots are a type of collaborative software-robot that work alongside human workers to improve productivity and efficiency in the workplace. They are designed to perform a wide range of tasks, from generating content and automating mundane tasks to providing feedback and suggestions to help workers improve their performance.

The typology of tasks:

Production-AI: This branch will be substituing things like content creation, whether textual or visual (still at ints infancy, but considering this is the most requested and consumed kind of content in today's web, and usually provided a big barrier to entry for many, has a much higher potential). With the rise of content marketing as a key driver of brand online strategies, creating high-quality content has become a critical aspect of many businesses. Co-bots can be programmed to generate blog posts, social media updates, and other types of content, or at least providing help on research, inspiration so that it can be handcrafted and improved later.

Examples: StableDiffusion, Midjourney (Visual), OpenAI, Jasper.ai (Content), Synthesia.io, Pictory (Video), Tome (Slide-making)

Automation-AI: One of the most significant benefits of co-bots is their ability to automate repetitive and time-consuming tasks, freeing up to focus on more complex and creative work. For example, co-bots can be used to generate reports, create presentations, and handle data analysis, allowing workers to spend more time on high-level tasks that require their unique skills and expertise.

Examples: SheetPlus.ai (Excel formula helper), Levity, Zapier’s OpenAI Integration (Worfklow automation), Obviously.ai, ChatGPT Code Interpreter Plugin (Data science without code)

Feedback-AI: Co-bots can also provide valuable feedback to help understand how to improve performance at work. For example, they can monitor and analyze productivity, identifying areas for improvement and suggesting ways to optimize workflows. This can help work more efficiently and effectively.

Examples: Inflection’s ‘Pi’ bot (General personal assistant), Socratic (Homework help), Youper (Emotional health assistance)

🌊 Learning: solving for abundance of information

Learning something new or mastering any art requires studying, and help from others. Whether it's people who have lived in the past and written about their experience, or someone doing a youtube video how-to, webinar, or helping you side-by side as an intern, colleague, mentor.

Back in 2010, when I was learning Digital Marketing, there was no MOOCs, Youtube was about cat videos, and there were probably 10% the number of online blogs about the practice that there are today.

"The abundance of information may paradoxically make it harder to find what you're looking for." Peter Singer

On the contrary, one of my biggest problems today when learning something new, is the abundance of information. I save and stack up loads of articles in my e-reader with an unhealthy FOMO.

Who do I trust? What do I prioritize? How do I remove the noise?

You either read-up all of the existing material, and make up your mind on your own, or identify someone you think has the expertise and credibility to do all of these things for you, and curate, summarize and explain.

Most of the times though, you don't get to talk to this expert. You read-up or consume visual content.

Co-bots provide a solution:

They have all the answers - By gobbling up all of the possible information they can get their hands into, Large Language Models (LLMs) can promise that you won't be missing out on anything - it's all there.

You interact with them - By understanding natural language, you are brought to think you can have a real-conversation with the software, and keep on asking until you get the answer you're looking for (also meaning sometimes they make this up, which is a real problem called 'Hallucination').

They actually show the work - One of the biggest problems when learning something new, is that everyone learns differently, and that you need practice and real examples, not just the theory. This is where LLMs make a big difference, and will increasingly make so even in the world of education.

📏 Personalization and regulation

While we’re witnessing the biggest arms-race for the internet since more than a decade, I’m thinking that there are some themes which should already be addressed in a strucutred manner (Chief Automation Officer, anyone?).

Personalization - Machine-learning models work best if they have a) a lot of data b) a lot of good, specific data c) training/iterations. This can be possible by feeding them with specific information, whether it’s your own or the company’s (obviously for different usage). Key question in this case is about privacy and intellectual property; bringing forward the case for Open-Source models, or a world where every company will have develeoped a proprietary version of this technology.

Regulation - We know technology moves faster than the law and organizations. The US government has just appointed Kamala Harris on regulating AI, but companies will need to build their own internal regulations to avoid confidential data leaks and unproper use, but also to control, predict and maximise the impact of this type of technology in the workplace.

Implementation - As mentioned at the beginning of the article, the first step should be to make an assessment of the 10% of displaced tasks, and build a plan of implementation that takes into consideration all of the factors we discussed above. A new architecture of work is coming, and the ones who will benefit the most will be those who experiment but start building guiderails right from the beginning.