Table of Contents
The History of AI
I wanted to sensationalise the headline for this article because that’s exactly what’s happening in the field of Artificial Intelligence (AI) & Machine Learning (ML) right now.
AI is not a new concept. In fact it’s been theorised about since the years BC. It wasn’t until the 1950’s when Alan Turing gave us ‘The Turing Test’ as a way of measuring machine intelligence that AI started to capture our imagination. We’ve all seen classics such as Terminator back in the day, so it’s hard not to be excited when you see examples of that come to life.
Fast-forward to the 90’s and the field of AI & ML saw major advancements. In 1997, IBM’s ‘Deep Blue’ machine beat the (then) chess world champion, Garry Kasparov. By the late 1990’s, AI was already starting to power Search Engines and other systems that would transform the World Wide Web.
Skip now to the current day and the likes of ChatGPT are making waves on social media and are being hailed as game changers. Like I said, this isn’t new technology, but the reason for ChatGPT’s explosive success comes in the form of their delivery.
For the first time, technology of this scale has been built in a way that the public can interact with it: text. They made it free, and they use an enormous language model to allow you to talk to it using the language that you understand. It then translates that into computer code, fetches the answer, then returns it back in the language that you understand.
Using the cost model from AWS, the cloud platform that ChatGPT is hosted on, it costs OpenAI roughly $3 million (USD) per month to run. This is based on current pricing where every query costs $0.0003 to calculate. Yes, that is very expensive, but they will eventually charge users to interact with it. However, they have a long way to go in refining the model before they can do that.
You Either Love or Hate AI
First of all, you’re allowed to be excited. All of this new technology being put in our hands is a lot of fun. But, it’s important to manage expectations. If you were hoping that ChatGPT would automate your business and make you an overnight millionaire, you probably still think drop-shipping from AliExpress is still the best kept secret in town.
It’s exciting as a concept because it appears to fit the criteria of a ‘get rich quick scheme’. Yes you can make money with AI, yes it can help optimise your processes and yes, it can also help you with ideation. But no, it can’t replace you (yet).
It’s also worth mentioning the category of people who actively dislike the idea of AI and are set on not using or ever planning to use such tools in their life and/or business.
If your concerns are over the ethics of AI, then you have a right to be concerned. There are a lot of discussions currently ongoing around the ethics of AI and have been for years now. The future of AI is downright scary by today’s standards and I’ll list my most likely predictions at the end of the article.
However, If you dislike AI because you don’t understand it, or you’re scared of it (not that many people would admit it outright), then I have an important message for you. I encourage you to learn about it, use it, even embrace it rather than put your head in the sand and hope it goes away. Whether you like it or not, it’s here, and it’s only going to get more powerful and dangerous.
The ChatGPT Honeytrap
You can ask AI to write an article or website content for you but it’s basically worthless if all you do is copy and paste.
Search Engines such as Google have stated that they consider AI-generated content as spam and they’ll score your website negatively as a result. You might be wondering how they can tell if it’s generated content or not, but it’s actually quite easy since tools such as ChatGPT uses a specific language model. Therefore, it’s easy to test content to see if it fits known language models. There are multiple tools online that you can use to check content yourself such as HuggingFace or GPTZero.
The other big issue is the quality of the content itself. Yes it can technically write you a factually-correct article all about a really niche subject but it still reads like a boring medical journal and lacks any creative flair or ability to stimulate the user into enjoying reading the article or website content that you’ve generated.
That’s why it’s not a get rich quick scheme, because you can’t expect to cheat having to put the effort in and expect to get the same, or even better results in return. You have got to stop treating your ‘customers’ as emotionless numbers. It’s so easy to lose the personal touch with people when you are putting them under blanket terms such as ‘customers’ or ‘orders’, especially if you aren’t meeting these people face to face. But those are real people behind your orders and they are highly emotional beings, driven by emotional, not soulless content.
Unrelated, but wanted to also quickly note that ChatGPT is also limited to content from 2021 and earlier. A LOT happens in a year and a lot happened in 2022 so whatever you’re asking it today, keep in mind, the answers are coming from over a year ago (at the time of writing this).
How to get the Best Results from ChatGPT
AI helps computers do things that would normally need your input. Things like learning, making decisions and solving problems. That’s the bottomline.
ChatGPT can help you write emails, legal documents, sales scripts, video scripts, the list is pretty much only limited by your ability to ask it the right questions.
The most success I’ve seen people have online, including myself, is to establish the format in which you interact with the tool:
- Establish context. A popular format for achieving this is by starting with “You’re going to act as…”. So for example, I would say “You’re going to act as an exercise and health expert”. ChatGPT is a language model so it needs to know what type of language it needs to be using.
- Provide the model with a task. Once you’ve established your context, give the model a clear task for it to complete. So if the prompt is about health and fitness, the task could be “Give the best advice when it comes to improving someone’s health”. Remember, this isn’t Google, you’re not looking for different websites, you’re establishing a conversation. Health is a large and complex discussion and you’re going to have a lot of questions so you’re really just setting the stage right now.
- Ask questions in your prompts. At the end of your input, you can say “Is that understood?” This helps ensure the output you receive is relevant and helpful by giving the model a better understanding of your request.
- Once you’ve asked your question, review the output and see if you’re happy with it. If it lacks details for instance, it’s time to refine your prompt. The more specific the prompt, the better the results will be. Ask additional questions and think about the context and tone of the conversation when you do.
Here’s how that interaction played out:
Other AI Tools
There are countless platforms and tools, now available but here’s my pick of the best ones to save you digging through them:
- Synthesia – Video generation
- Lexica – Find images you like, and see what prompts were used to create them so you can learn and use those prompts on platforms such as DALL-E & Midjourney.
- Wellsaidlabs – Text to audio speech
- Runway – Video editing with AI tools
- ChatGPT – The current king
- Leiapix – Convert your 2D images into 3D
- Scenario.gg – Take a particular art piece and create a heap of variations from it
- DALL-E – Image creation tool
- Midjourney – Image creation tool
The Ethics of AI Content
The tools you use to create your content, whether it be text, image or other, all need to come from somewhere. Generative Artificial Intelligences (GAIs) such as DALL-E and Midjourney are all trained on existing content. That content is often pulled from the internet, from us. Is it ethical? Is it even legal then that these AI tools are creating content, using someone else’s work without consent or compensation?
In 2018, Google announced that its smart reply feature had been trained on “billions of Gmail messages”. So, that edgy Facebook post that you made 6 years ago very well may have been used to train someone’s language model.
All Machine Learning models have this issue and based on the results so far, no one cares about privacy or copyright at the moment. The short answer from the companies creating these tools are “so long as the material itself was lawfully acquired and the model doesn’t plagiarise the material” then we can all look the other way. This mentality, is dark grey at best. I wouldn’t be surprised to see more strict rules around this technology in the future.
As for your own legal rights, all major models give you, the end user, rights to commercialise the content generated by their prompts. So the ethics really come down to the individual user and their own moral compass as to how far down the AI rabbit hole they’re willing to go.
Future Prediction for AI
In 2018, Google showcased their Google (AI) technology (though it was never released to the public). During the demonstration, the AI had a real-time phone call with a human and booked a hairdresser appointment for a “client”. They discussed the dates and times available and it was indistinguishable from a real human to human conversation.
Now imagine it’s a sales call. The sales AI has listened to millions of sales calls and knows exactly what to say in any situation. This sales AI will have 1000’s of lifetimes more knowledge and experience than even the best human salesperson on the planet.
This sales AI can have sales calls 24 hours a day, it never needs to take a break, never gets sick and it can take calls all over the world. This AI expert will exist across any industry or profession where the trade is information.
How about a medical AI in your own home that can provide an immediate, and more accurate diagnosis than your physician.
Or a finance AI that can take swarms of raw data, and realign it into simple-to-read excel sheets for you.
Superintelligence becomes a very real discussion if you are to imagine what version 100 of ChatGPT looks like, and combine it with version 100 of Tesla’s bot, Optimus. Machines will look like us, talk like us and most importantly, be smarter than us.
Experts around the world are placing bets on when they think the Singularity will occur (the moment in time when artificial intelligence becomes smarter than humans). 2045 – 2050 is a popular time frame for a lot of experts, with some saying even earlier. Suddenly the concept of Skynet doesn’t seem that wild as a film plot anymore!
The Takeaway Lesson from all of this
Repeat after me, “AI isn’t here to replace me, it’s here to empower me”. If you can’t adopt this mentality, you may as well retire now, before things really kick off. Embrace change, it’s part of humanity, technology moves, we have to adapt.
Think of the number of daily tasks that require your input. Emails, Slack messages and other notifications. With AI & ML, you could easily tell your AI to predict what you are going to say and write those inputs. Then all you need to do is validate the list and you’ve suddenly done a day’s worth of work in a few minutes.
On a more professional level, you can use AI to help you with the content generation aspect of your job, leaving you to QA the work. This gives you a higher output, allowing you to make more money or become more valuable to the company or client you are working for.