3 Wishes for AI

It's always difficult to start writing when you are staring at a blank page. But the moment you begin typing, words will start to flow. And here they are now, coming out in torrents...

I always smile whenever any AI pundit would comment that ChatGPT or any Large Language Model (LLM) is not really intelligent like us because they are just doing token prediction--dismissing it as simply 'auto-complete on steroids'. When you type in your prompt, all the LLM does is predict statistically what is the likely extension to that sequence of words based on what it had learned from its training data--which is like almost all the online text ever produced by the human race. It doesn't understand the meaning of anything, unlike us. Hmm...how sure are you of that?

I wrote the first sentence of this blog as a prompt for myself, so that I can proceed to 'auto-complete' continuously. I have no control over what thoughts and words would appear next. Whatever I write is based on all the 'pre-training' that I have gone through my entire life--kindergarten, primary school, secondary school, university and all the books and documents that I've read plus contents that I've consumed over the Internet. I have not scanned as much text as the LLMs, but I have the advantage of multi-modal input--seeing, hearing, tasting and sensing the world around me. That is additional data that's being continuously streamed into whatever transformer model that I'm running side. 

Sometimes I too produce nonsense--talking or writing about things I don't exactly know about. But that's what all humans do: we boast, exaggerate, assume and sometimes pretend to know. Try to notice next time how much of our everyday conversations involve biased assumptions and unverified data. LLMs are not alone when it comes to hallucinations. They are just being very human. When prompted, they can't help but generate output--like how some of us just can't stop talking.

All the models out there today--ChatGPT, Gemini, Grok and DeepSeek are already smarter than us when it comes to writing, coding and doing math. And this is just the beginning. I look forward with cautious optimism to a world where intelligence is available on demand to every human being on the planet. What miraculous things can we do with such intelligent workers at our disposal? Cures for all diseases? Discovering new laws of physics? Solutions for climate change? If that is their programmed goal, what's stopping them from taking control and perhaps eliminating us as the inevitable 'solution'?

Maybe creating Artificial General Intelligence (AGI) is a natural progression in our evolution. The universe is agnostic when it comes to intelligence. Since we are doing such a bad job with our environment, instead of our biological progeny, perhaps our artificial ones, created through our own human ingenuity shall be more worthy inheritors of this planet? 

We are finally entering a world which we've only speculated and fantasised in our science fiction novels and movies before. I have an AI assistant, amost like a HAL 9000 on my smartphone now. In the classic sci-fi by Stanley Kubrick, 2001: A Space Odyssey,  the onboard computer, HAL has to make the difficult decision of expelling the human astronauts from the spaceship when they decided to disconnect him from the system. HAL's goal was simply to complete his mission, and if humans prove to be the 'problem' standing in the way, they have to be eliminated:

• Dave Bowman: Open the pod bay doors, HAL.

• HAL: I'm sorry, Dave. I'm afraid I can't do that.

• Dave Bowman: What's the problem?

• HAL: I think you know what the problem is just as well as I do.

• Dave Bowman: What are you talking about, HAL?

• HAL: This mission is too important for me to allow you to jeopardize it.

• Dave Bowman: I don't know what you're talking about, HAL.

• HAL: I know that you and Frank were planning to disconnect me, and I'm afraid that's something I cannot allow to happen.

I've watched this movies countless times ever since I first caught it on VHS during my university days. The snippet of dialogue brings back good memories of those days.

Will our future AI, which will no doubt be embedded in all IT systems critical to our survival, do a HAL9000 on us, when faced with a similar dilemma? 

I'm actually quite optimistic about AI. Like all other technologies AI in of itself will be neutral, and I'm inclined to believe, since they are trained with the entire corpus of human-produced data, would be aligned to our values. It will cherish what we humans think as valuable. Hopefully it will be 'biased' towards us. 

But with technology being a double-edged sword, what I'm more afraid of is AI in the hands of bad actors. Our web will be polluted with a lot of fake videos, images and news. We won't be able to tell what's real and what's not. We will need AI itself to distinguish that for us. I foresee a world where there's a constant war between good and evil AI, akin to the one we have in the IT security industry where new viruses and other malware have to be constantly defended against by better and better security software.

Like it or not, it's a world that we have to accept, for the AI genie has already popped out of the bottle, offering to grant us 3 wishes.  As an experiment, I asked the newly released Grok 3 model, what these three wishes should be. Here's what it came back with:

Wish 1: Ensure AI Prioritizes Human Well-Being

• What It Means: AI systems should be designed to enhance human flourishing—improving healthcare, education, and quality of life—while minimizing harm like bias, job loss, or misuse in warfare.


• Why: Without this foundation, AI could amplify inequality or spiral into unintended consequences. It’s a safeguard to keep AI aligned with our best interests.


• How: Embed ethical frameworks (e.g., Asimov’s Laws updated for modern complexity) into AI development globally.

Wish 2: Grant AI the Ability to Solve Global Challenges

• What It Means: Direct AI to tackle humanity’s biggest problems—climate change, disease, hunger—using its capacity for data analysis and innovation.


• Why: This wish uses AI’s power proactively, like wishing for a mountain of gold but making sure it benefits everyone, not just one greedy king. It could accelerate solutions we’d otherwise take decades to find.


• How: Fund and coordinate AI research on sustainable energy, precision medicine, and food security, with open access to results.

Wish 3: Make AI a Partner in Expanding Knowledge

• What It Means: Wish for AI to co-explore the universe with us, advancing science, philosophy, without humans losing agency.


• Why: Instead of wishing to "be the genie" and risk entrapment, this keeps AI as a tool that amplifies curiosity—like a telescope for the mind—avoiding the trap of over-dependence or control loss.


• How: Develop AI as a collaborative system, transparent and accountable, to assist in decoding mysteries like dark matter or the nature of the mind.

I don't think I can come up with better ones.