This article is the first in a series on Artificial Intelligence, each addressing increasingly complex concepts of AI and its human collaboration potential. The million-dollar question is whether Artificial Intelligence will remove jobs or enhance human-machine interfaces. We shall see in this article.
Table of Contents
How about starting with Human Intelligence?
So, what makes humans Humans? There are two main aspects to them: Cognition and Intelligence. Yes, they are different! Now, as long as artificial intelligence does not replicate these two aspects in full, it cannot be true AI. Ok! So what are they?
Human Cognition
It refers to the background processing that the mind does 24×7, and they are:
- Acquiring knowledge
- Processing it
- Storing information (Experiences)
- Using this information (Applying)
These steps encompass all other basic activities of our human mind, which are:
- Perception: Interpret what our 5 senses gather and provide as inputs to the brain.
- A person entering new surroundings, say, an automobile showroom, becomes aware of everything there, including the salesperson, the types of automobiles on display, their colours and shapes, etc.
- Attention: Focus on what is needed and filter out the rest.
- In a discussion on an interesting topic, attention is required to participate. It helps us focus by filtering distractions like noise or our chattering minds!
- Memory: Storing of information for later recalling.
- We need it to remember where our home and office are, where we have parked our cards, etc.
- Learning: obtaining and using fresh knowledge and skills.
- Learning a new language is a typical example. Continuous exposure to vocabulary and grammar will reinforce our ability to converse in that language.
- Reasoning: Logical reasoning and conclusions helped by past experiences.
- When planning a trip abroad, we need to consider various parameters like location to visit, why, and at what cost. We can only develop a reasonable one if we have reasoned properly.
- Problem-Solving: Coming up with appropriate outcomes in situations.
- A typical example is solving a math problem from its problem statement. When your car breaks down in the middle of nowhere, your approach to trying to troubleshoot, and if you cannot find one, the ability to decide when to call the breakdown service number, is also problem-solving.
- Decision Making: Judging to choose the right from several options.
- When buying a home, one must consider several options and pros and cons before deciding.
Human Intelligence
Human Intelligence refers to using all or some of the cognitive processes listed above to solve problems, adjust to changing or new situations, and obtain results. Intelligence is a subset of Cognition. We find each individual is good at something or the other. Each person’s intelligence is different. This is attributed to their ability to use a particular cognitive process.
Based on the dominant cognitive process used, intelligence can be categorized into the following types:
- Analytical Intelligence
- When an individual’s problem-solving and reasoning abilities (cognitive processes) are higher than others, they can analyze and differentiate information.
- Creative Intelligence
- A person’s ability to use their imagination and idea generation (cognitive processes) to generate new ideas and outcomes on actions is the creative intelligence that is at its best in that person.
- Emotional & Social Intelligence
- This type of intelligence uses perception, self-awareness, reasoning, problem-solving, and attention. It uses many cognitive processes and is, hence, more complex than the other types of intelligence.
We see that the human mind is a complex entity, and many aspects are still being researched. By digging deeper and deeper, scientists are gaining new insights into it. The Artificial Intelligence machines or engines (the programs/algorithms) that run are invariably super complex and obviously need to be dynamic if not as much as our brain, but at least have the ability to constantly keep adding to its learnings.
So what is Artificial Intelligence, then?
This is a very serious attempt by mankind at the top of all cutting-edge research and technologies of the current world. The artificial intelligence engines are a bunch of programs executing algorithms that try to simulate a combination of human Cogniscence and Intelligence processes. The core of an Artificial Intelligence engine, at its barest minimum, is a combination of these:
- Data
- Algorithms (software programs)
- Power of Computing
The ability to replicate or simulate the human mind has been challenging, and progress has been rapid. We will try and capture the gist of these stages of artificial intelligence growth:
Automation & Machine Learning (ML)
We encounter automation quite often in our daily lives. These machines generally replace humans’ repetitive work and make them relatively independent of human intervention. Conversely, ML enhances automation by adding the value of adaptability, protective nature, and pattern recognition. While ML involves some cognitive processes like prediction and pattern recognition, automation, as a stand-alone, only involves data comparison with previously stored data and logic (yes/no or right/wrong).
Example: While browsing Netflix or Amazon, pop-ups may recommend related content. These are examples of ML and automation working together and refining constantly to provide tailored recommendations.
Deep Learning (Neural networks)
These are areas of artificial intelligence that are more complex and closer to human cognition and intelligence. NN works similarly to the human brain, interconnecting millions of neurons when triggered. It processes data in layers, each improving the change of a hit using biases and weightings.
Example: Assuming that the artificial intelligence engine is tasked with recognizing the number ‘9’ provided in pixel values (784), the algorithm processes this number at various layers, each using biases and weighting given to 9 numbers (0-9). Finally, it comes up with a maximum probability for 9 (input data under test), while other numbers get lower or insignificant probabilities. The picture represents a neural network maize having 3 layers of neural network.
Collaboration of Human-AI is the way forward!
All said and done. It may sound like Artificial Intelligence is all set to overtake humans; read on to know more. In today’s scenario, cyber security is a major domain, and its role in Security Operations Centers is limited to flagging threats, and human analysts take the final call. The main reason has been found to be the confidence in the accuracy of the data that drives AI engines. Until accurate data is not available to these engines, humans in the loop is essential.
Noting that AI can act as a double-edged sword since criminals can also use AI to support their nefarious activities, it becomes all the more necessary for humans to be monitoring an AI engine to press the ‘ABORT’ button! Because of these limitations and evolving phase of Artificial Intelligence tech, a prudent approach would be to have systems that Collaborate with humans. Here are a few pointers to foster this confluence:
Quality of data and Engine training: Priority on quality data and regular engine training will enable robust AI systems that can ward off ‘sleeping attackers’.
Protect login credentials in AI systems: once they have the credentials, cybercriminals can easily enter and stay dormant and move laterally and steal information in stages without flagging security breaches.
A tiered approach to AI implementation: If AI flags a usage pattern as a risk, humans can intervene to assess it as a potential risk.
Protection Identities: Make identity security a thing all employees understand, irrespective of their technical capability. This can be done by having instructions in plain English.
Coming in PART 2!
Natural Language Processing (NLP)
Perceptual AI (Computer Vision)
Adaptive Learning
More about these in the next parts. We stop here and step back to see what we have understood. The reader is encouraged to refer to other sources on Basic AI and Deep Learning to reinforce the learning here.
Brilliant
