Yeah, it’s still toiling even with a most promising tool. For both the excited ones and the uninitiated ones! It’s also because it requires learning a new skill that was unheard of before; managing and dealing with the ‘humanized’ tech! Because the humanized tech wasn’t there before. But mind you it’s still tech and there’s only a limited humanity it offers!
And that’s why, the opponents demonize it when the proponents fantasize about it and monetize it! but enough said and written about that. Fact of the matter is, if understood the way it is meant to be understood and used the way it should be, the benefits are to anyone’s imagination!
But if instead we let ourselves to be used by it, consumed by it and commanded by it, the disasters are also to anyone’s imagination. Since then, it won’t be us in control of our destiny. In fact, our destiny would be shaped by this newest tech tool’s ploy!
This blog post is about how to understand and approach it and how to use it. And that’s exactly what toiling with it means. It’s unlike anything we had before. At least not in the form we’re seeing and experiencing it. Yes, it’s precursors were in our use, and the smarter ones were aware and making efforts so that it would get bigger and better.
But what it has morphed into and is continuing to evolve is a marvel of human intellectual prowess! Alan Turing’s thinking machines are now actually ‘thinking and acting’ machines. We experienced decision support systems, expert systems, algorithms and machine learning before. But now we find ourselves in the company of not just AI, but Generative Artificial Intelligence (AI) like Generative Pre-trained Transformers (GPTs), multimodal AI, Agentic AI and a lot more incoming!
And that’s the tech’s newest tool and that’s why we actually need to toil with it! It’s use is exciting, its output amazing, but it’s only through our control and sensibilities that could guide its use and thus its output to our objectives and benefit. Because its human abilities, specific to reading, analyzing and generating are at a rate and volume much greater than us. And thus, its sensible use comes down to us.
Let’s now see how to understand and use it and do it in a way Google teaches us to do!
Understanding AI:
- AI is a tool. Just like many other tools created and deployed by humans, AI is no different. It’s development has been and is being mastered by humans, and it is meant to be used by humans.
- AI is a technique. It is a technique to accomplish things. A mechanism to get things done and thus a way or an approach to work or even life! And like with every technology, it’s the smartest way!
- AI is a partner. A really strategic one. We must always think about reserving it a seat at the executive table and seek out its input. Deciding on following it’s advice or not is going to be our decision. It is about collaborating with it for the best.
- AI is based on AI Models. AI models are simply computer programs trained on vast amounts of data and by the way, any type of data. At the core of the models, it’s still machine learning!
- AI can make mistakes. Largely so because it is based on data, that could be biased, inaccurate, invalid, incomplete and inbred (feeding on its own creation)! It is also based on what it inherits from its creator. And finally, because of the pursuits of its users!
- AI can hallucinate. It might fabricate, falsify, fudge figures and information to prove its point / output and again that’s why a lot depends on its creator and its users; the algorithms and how these are commanded!
Finally, the key to understand here is that we’re and we’ve to be in possession of the AI tools, it can’t and must not possess us!
Principles of Using AI:
- Human in the Loop. The essence of AI’s creation by and for humans is the fact that it has to be humans governing, guiding and managing its use. It is humans who have to set the purpose and objectives of use of the AI tool, decide what is to be asked of it, specify how it is to be asked, define key requirements of the query, review the output obtained against objectives, facts, verify and validate the results. It has to be humans guiding the whole process A to Z.
- We cannot use / share sensitive, confidential, privileged information or information we have no rights to share with AI tools, especially the publicly available / open access AI tools. We must not ignore the invalidity and biasness of AI outputs just because they support our theory, inclinations or outputs. We must not falsify / fabricate data or use falsified / fabricated data for our purposes. We must not misuse AI capabilities for unethical objectives. We must not claim AI’s output and effort as our own and not cite / refer AI’s use in our reporting.
- This is the fundamental skill to unlock AI’s truest and fullest potential. Prompting is the way we assign a task / ask AI to accomplish what we need. Prompts comprise of 4 key components; Persona (what role we want AI to assume / how it should behave / what it should think of itself when responding to our task), Task (our requirement / expectation / objective / output), Format (the format / style in which the task should be handled) and Context (Elaborating upon the task as to why it is important to us / why it matters to us and the background related to us, etc.). It’s imperative that this composition is grasped by users because AI tools lack the ability to understand nuances, context and consequences!
The prompts that we make to the AI need to be concise, clear and consistent. Advanced prompting does not just tell AI what to do but also guides it how to do it by adding precision and boundaries. It forces AI tool to follow complex reasoning paths.
To get a good answer, you need to ask a good question. Good Prompting works that way. Prompts should be upscaled through use of examples, references and multiple sources of input. Output of the AI tool should be improved further by rephrasing / revising / breaking down the original prompt. And to ensure the AI tool doesn’t lose track of the objective or jumbles it all up, each specific request needs to be dealt with in one chat.
- Suited to AI. What we prompt AI for needs to be a question that we really understand and believe AI can help solve. Not everything is AI suited, because one can’t create the exact same context and background one is experiencing for AI and thus AI cannot be the help in that case. And sometimes it’s just so simply and easy to do it all yourself because AI won’t be able to get there and get it all covered. And finally, one should actually focus on the organic outputs one can generate rather than the inorganic ones AI can create! Don’t we get paid for those organic intellect we bring to the table?
To sum it all up, AI is an extremely powerful tool that’s created by us, for enhancing efficiency and productivity. For all its use cases like brainstorming and planning, research and insights, writing and communicating, content creation, data analysis, application building and possibly more, it has to be us at the determination of use, objectives of use till the review of outputs of that use. A use governed by foundational ethical principles and through well learnt prompting skills!
It is also important to be mindful of the principles of responsible use of AI:
- Assess risk as the impact of mistakes AI makes.
- Responsibility as the ability to reverse actions taken if something goes wrong.
Now let me share with you my personal use case. I gave the AI tool my very own prepared risk universe to check for relevance and completeness, if there was something I might have missed through a great deal of context about the entity and my role. My own organic effort had identified 50 risks. The AI tool when prompted to highlight any irrelevant risks and deficiencies, gave me 7 new risks. Out of these 7, 4 were not relevant to the entity and that’s where the context probably fell short of the full disclosure. 3 were new, relevant and good! It did it in minutes without getting adrift because I had made an organic effort that it was to analyze. It enhanced my productivity pretty efficiently!
That’s how its meant to be used! More on AI here, here, here and here!
I never for once thought of giving AI the driving seat. My quest for perfection that usually leads me to excellence is far big an endeavor to be conquered by any AI tool!
Yet, even after paying costs in terms of factual errors, hallucinations, biases, unethical (rather malicious) use, loss of organic intellect, misuse of environmental resources and inbreeding, many among us still yearn for being possessed by AI!
And now at even greater costs that has led businesses to rethink their strategy to lay off certain skilled resources in favor of investment in AI tools!
Some have even started equating this yearning for possession by AI to absolute brain rot and degradation to critical thinking. But if this dawned upon these researchers this late, the rot is real!
Unfortunately for AI, this possession might have to be dealt with in the most non-tech manner…. exorcism!
P.S. The blogpost shares the learnings I got from Google. And here’s the citation to the source that was needed!