An example of a Predictive machine in action: if they drop a glass of milk from a table that is 1 meter high, a Muggle will know the next step is it will probably fall to the flaw because of gravity, and then the glass will break into at least five three pieces. These are mathematical and scientific high probability effects supported by similar and concurring data from past similar situations, and therefore that is what a Predictive machines result will be.

Generative machines (like Mr/Ms Chat GPT), create content, such as text, images or sound that resemble human-created content. The information (data and scenarios) is processed through algorithms but the ‘result’ is presented in a way that mimics a human character like outcome or action.

For example, if a Generative machine drops that glass, in addition to knowing the glass will break, it will blurt out a swearword and then call for his wife to help clean up the mess. This is what we might expect a human to do and that is what sets the two types mentioned above so far apart from one another.

There are members with capabilities and categories of functions that cross over in ‘form’ making it impossible to define an AI perfectly, just like a human, and who knows, it might want to ‘identify’ itself as another type later, but for the purpose of this AI Manual, let me try regardless.

Prediction and Generative Machines, are Muggles!

Muggles or Predictive machines are simplistic clustering ‘machines’ which group similar data points based on certain characteristics and identifying patterns. Predictive machines can then make a prediction based on what has happened in similar past (existing similar data) situations by proposing the next ‘step’, and this is in most cases determined by probability using an algorithm. Did I lose you there? Stick with me just a moment longer and I promise you will understand perfectly.

An example of a Predictive machine in action: if they drop a glass of milk from a table that is 1 meter high, a Muggle will know the next step is it will probably fall to the flaw because of gravity, and then the glass will break into at least five three pieces. These are mathematical and scientific high probability effects supported by similar and concurring data from past similar situations, and therefore that is what a Predictive machines result will be.

Generative machines (like Mr/Ms Chat GPT), create content, such as text, images or sound that resemble human-created content. The information (data and scenarios) is processed through algorithms but the ‘result’ is presented in a way that mimics a human character like outcome or action.

For example, if a Generative machine drops that glass, in addition to knowing the glass will break, it will blurt out a swearword and then call for his wife to help clean up the mess. This is what we might expect a human to do and that is what sets the two types mentioned above so far apart from one another.

Let’s move onto Hufflepuffs or Narrow AI. This is the most ‘advance’ AI available today. Artificial narrow intelligence (ANI) or weak AI, describes AI designed to execute specific commands only, although these commands or ‘prompts’ can be extensive. ANIs are built to do well in one rational type of capability, but cannot independently learn new skills beyond its original design (coding).

Natural language processing AI is a type of narrow AI. It can identify and respond to voice or text commands (human ‘like’ data interfacing), but cannot perform tasks beyond that. It cannot write new code or algorithms or decide it wants to become a coffee machine instead.

Siri, Alexa and Google Assistant are all Hufflepuffs. Hufflepuffs have access to a far greater data set and can more easily engage with us in a human like manner.

If the above has confused you a little, that’s not a problem. All you need to remember is that the AI family members mentioned so far are keen to learn more but have limitations, and are for the most part socially awkward.

Let’s move onto family members who challenge the norm but do not yet exist. The Slytherins: Artificial General Intelligence (AGI) or Strong AI. These represent family members that can learn, think and perform a wide range of actions outside of their predefined (programmed) and assigned characteristics, functions and tasks.

For example: a man deciding not to go to work because he would rather stay at home and raise the children is acting like a Slytherin by challenging the social norm.  

The goal of designing Artificial General Intelligence is to be able to create an AI capable of performing multifunctional tasks that far exceed its original design. This AI will have a unique reason for its own decision, and will decide on which data to use and how to process that data. It will pick what it wants its outcome to be and work towards that result.

For example: if we programmed and provided data to an Artificial General Intelligence Robot that most meals are consumed using a knife and fork, but then gave it a bowl of soup, it would stop and think about the situation before trying to eat. This sets it apart from Hufflepuffs and Muggles who will still try to gobble up the soup with a fork. AGI Slytherins will search for alternative data that might work better in this new situation until it finds a spoon in the database.

A Slytherin is also the type of AI that would blame the cat for knocking the glass of milk off the table (even though it didn’t) in the prior scenario, because it will have a basic level of self-preservation and perhaps even ego. When we eventually make something this intelligent it will start to propose things like the world might be more efficient without certain life forms and ask: “Why do we need to run the risk of having the glass fall off the table at all? Why not remove the ‘thing’ that needed it in the first place?”

Artificial superintelligence (ASI), could be the Harry Potter or the Voldemort of AI. ASI is a hypothetical intelligence way beyond the scope of human comprehension. It will be built upon the foundation of Artificial General Intelligence, Supercomputers, Quantum hardware and Generative AI models, but probably then go on to build its own new hardware and write new code for itself.

Once ASI is ‘conceived’ I believe that in a matter of days (perhaps even hours) it will have the thinking capabilities of humankind’s collective brainpower and knowledge.

ASI is set apart from other Slytherins because it will be self-aware, set its own moral compass, replicate and expand itself in physical form. This scenario is referred to as the Singularity, and ‘this moment’ has been depicted in many movies and books.