The Reality of AI Hype and The Future of Business

Welcome to edition #20 of the “No Longer a Nincompoop with Nofil” newsletter.

 Subscribe 🖌️ | Go Premium 🚀 | Work with me 🤝 | Discord 🤖

Here’s the tea ☕

• The ChatGPT hype is dying 📉

• The future of agriculture 🚜

• Worker shortages are a thing of the past 🤖

• Multi-modal models are getting good 🧬

How long will the hype last

It’s been just over 7 months since ChatGPT was released and it is genuinely hard to imagine the world without it. So much has happened in the AI space since then; calling it a technological revolution would be an apt description. We’ve seen more innovation in the past half a year than we have in the past half a decade. It’s not that ChatGPT has been the cause, but rather the catalyst. The world woke up to what LLMs could do, and there is still so much we’re figuring out.

You might have seen this graph showing how fast ChatGPT got to 100M users. It essentially became the fastest growing consumer product ever, doing it in just 2 months, an insane achievement. But is this growth sustainable? Is ChatGPT, as it is right now, a product you use on a daily basis? Is it something people would use daily?

Naturally, there is no blanket answer here. For some people, ChatGPT is a godsend. It can write code, emails, blogs, articles, teach you anything, discuss any topic with you; it can do a lot. But when does the novelty wear off? What happens when people need a specific tool for their specific use case. A tool that can’t be wrong, even once. Something they can rely on without having to worry about hallucinations. Well, we might be looking at this exact scenario playing out.

For the first time in months, ChatGPT declined in usage. Now, there could be any number of factors this can be attributed to. For example, schools might be on holidays so students have no need to use it at this point in time. Although this very well might be the reason, I think this represents an inevitable future of ChatGPT and AI in general. Besides the general fun use cases that people can play around with, AI is best used in situations where there is specific industry application. What does this look like? Let me explain.

Over the last few months I’ve spent time meeting with dozens of industry experts, founders, CEOs and business owners exploring how AI can help them with their work. At this stage, what I’ve realised is that there is still a severe lack of understanding regarding what AI can really do; how powerful it can be in transforming not only businesses, but entire industries. The real power of AI will be seen when people with knowledge of an industry will work with people with knowledge of AI to transform the world. I don’t mean to exaggerate when I say transform the world either.

The use cases for AI in disrupting every single industry are practically limitless. I’m lucky enough to say I’ve seen first hand how real the disruption will be; not only from all the things I’ve read and seen, but the conversations and consults I’ve been doing as well as the products I’m helping companies build.

The value that will be created with AI will change the way the world works forever. This won’t come from simple tools people are building over a weekend. It will come when companies take an AI-first approach to their entire existence. When every company will have an internal suite of AI tools powering their operations. The smart companies are already doing this, and funnily enough, selling it to their own competitors. Funny how that works.

The future of agriculture

This is the LaserWeeder by Carbon Robotics. It is a fully autonomous, AI-powered, chemical-free weed killer. Zero human intervention required. It runs on Nvidia GPUs and uses 42 high resolution cameras to identify and target weeds. It’s been trained on over 40 different deep learning models and its high powered lasers can kill over 200,000 weeds an hour with sub-millimeter accuracy. In fact, it’s so accurate it targets the apical meristem of a weed, rendering it unrecoverable.

This machine runs day and night, sunshine or rain; it doesn’t get tired, it doesn’t disturb the soil and it doesn’t require herbicides or pesticides. Sure it might carve a hole in ones wallet, but in the long run, it will pay itself off relatively quickly, especially considering labour is one of the biggest costs for farmers. It is a realistic and viable way for organic farming, helping us live healthier lives.

This is the future of agriculture, and it’s powered by AI.

The end of the worker shortage

The agriculture industry has been facing a worker shortage for some time now. Many places rely on immigration to keep up with laborious jobs on farms. During Covid, since travel was restricted, Australian farmers had a very hard time finding workers to pick fruit. Well, that’s a problem we won’t have in the future.

The system uses the drones to pick a range of fruit. It analyses the fruits sugar content using various image recognition models and gives farmers real-time insights into their harvests.

I do want to emphasise one thing here - this is not something that couldn’t be done 5 years ago. In fact, it’s hardly being done even now. The problem does not lie in the feasibility, but rather it takes time and money to research and test and approve such systems. Unfortunately, this is just the tip of the iceberg. Issues like maintenance and customer acquisition are probably the biggest barriers for these type companies. Unfortunately, potential buyers are generally not tech savvy either.

Taking all of this into account, you start to understand why technology like this isn’t already mainstream. The capability is there. The first company to properly build, market and scale AI systems for agriculture will an incredibly profitable and influential enterprise. I’m looking forward to seeing it happen.

What an AI wearable would look like

Otter is an open-source multi-modal model that can analyse real-time events and provide feedback on how to tackle them. The team from NTU in Singapore have used OpenFlamingo, an open-sourced version of Google DeepMind’s Flamingo model. Google’s been using this model to watch uploaded Youtube Shorts videos and create tags and metadata for them.

Compared to Humane’s wearable device, this is more like Google Glass. What exactly a wearable AI device will be like is still unclear but the potential is very clear and we already have models that can be used in real world scenarios.

As always, Thanks for reading ❤️

Written by a human named Nofil