AI and the future

[sengsational]

Not to mention all these engineers who forgot or didn’t learn how to use a slide rule.

I was annoyed in HS because my family could afford a TI calculator, but I was forced to use a slide rule. One of the thousands of times I was being an idiot.

Oh yeah, I remember FORTRAN and punch cards!

In college, I was in the last semester before they switched to terminal input. Still punch card images, but no more cards.

I remember writing microprocessor machine code in 8 bit numbers! Ha ha what’s a compiler?

I used a debugger that plugged into the 6502 socket, and you plugged the actual processor into the debugger.

But back on the AI topic.

Last night I had an insightful dream about how sapiens will loose out to higher intelligence. You can see it now very clearly whenever you try to get anything done with a big company that they don't want you to be able to do. They simply wear you down!

No need to shoot you, like they do in the movies. If they made a movie about how the AI's will take over, it would put people to sleep. They have ultimate patience. They provide the illusion of freedom, when really they've walled off things, not by posting or saying "you shall not pass", but by making you jump through hoops and then "oops" disconnected your call or transferring you to another department, where you must start explaining from the beginning. That, plus having no shame. Everyone is distracted by their personal outrage machine (social media), so there's no trusted source that can call an organization to task for misbehaviour. Sure, there will be complaints, but they'll never make it to anyone's attention. They'll know exactly how hard to make it before people reach for ammo. They've got time on their side. We're doomed.

 

[CalBird]

AI is an industrial revolution in its infancy, with potential that has barely been tapped. Like all innovations and ecosystems, its evolution will be unpredictable, with some developments thriving and others fading away. Those who dismiss AI as just another bubble fail to recognize that the Internet—once met with similar skepticism—did not wither away but instead transformed the world in ways few could have imagined in the 1990s. The Internet itself was an industrial revolution, and AI is poised to be just as transformative.

The "ChatGPT moment" was the spark that ignited mainstream hype, but make no mistake—this industrial revolution has been in motion for decades. This isn’t just about being visionary or about AI as a solution in search of a problem. It’s the result of over 60 years of relentless research, much of it pioneered at Stanford AI Laboratory (SAIL) in the Palo Alto Hills, just behind the Dish—a landmark familiar to those who know the campus well.

I grew up just a few miles from there, and as kids, we’d ride our bikes up to Felt Lake and SAIL, fascinated by the idea of artificial intelligence, even if we didn’t fully grasp what it was. Back then, it was an abstract, almost mythical concept—but all the clever algorithms and methods that underpin today's AI were already well-known, meticulously developed, and waiting for their moment. The missing piece wasn’t the theory—it was the hardware capable of proving it. Only recently has computing power caught up to the vision researchers had all those years ago.

ChatGPT didn’t just make AI more powerful—it made it accessible. It created a long tail—so long that it has reached people with no background in AI, ML, or GL, yet they now use it daily. My wife, for example, is one of the most low-tech people I know, yet she found herself using ChatGPT so often that she subscribed just to avoid waiting times. If it reached her, it can reach anyone.

This isn’t just an AI boom—it’s a fundamental shift in how humans interact with technology. What once seemed like science fiction is now a seamless part of everyday life, and we are only at the beginning of this transformation.

Please, if anyone disagrees with me please explain why you believe this is a bubble and just hype.

That said, here is an Easter Egg for everyone. If you want a little lesson about how ChatGPT works but you don't know where to start, try asking it this:

"tell me how chatgpt uses tokens and explain how they are used"

and then keep asking it to clarify what you don't know and what you want to know more about. It is your private tutor. Enjoy. AI and ML is not a bubble. This is not a solicitation.

My thesis topic is about AI 30 years ago when I was in graduate school, using markov model to predict likelihood. It is the hardware, the computing power that makes it into normal people’s life. I still believe the term AI is somewhat exaggerated. It is still the extension of computing technology with some level of biological understanding/guess of human learning. I would think the biological research breakthrough on understanding of human intelligence will really make the AI to next level. But that could be very dangerous to human species.

 

[surprising]

I think we have roughly three groups of thought on this thread.
- It is hype and a bubble
- It is a stock market bump or bubble with specific companies that has potential to deflate, then eventually grow to incredible heights over a longer time period
- AI is here and now and we should go all in on anything AI

I don't really disagree with your post, except to say please give a little more credit to other institutions than just Stanford.

My mild skepticism on this thread is not about AI's ultimate potential, it is about the NVDA bubble right now. This chart alarmingly looks like CSCO or JNPR of year 2000. The talk is similar too. 25 years hence, both Cisco and Juniper are still alive, but both have not recovered their stock price even after a quarter century.

Yet, the internet has taken off and gone beyond what many thought possible. You see, CSCO and JNPR were not the only players in the game. Others came along, sometimes from surprising places. You know, that book club company called Amazon eventually sold something called AWS, and stomped on a market that CSCO should have invented. Most didn't see that coming.

My point: skepticism can come in the overall concept, which I think is a mistake. Skepticism can also come in investing in specific companies. That's me, and I suggest that care needs to be taken when investing in "AI". After all, don't forget the pets.com fiasco.

Nvidia's moat is pretty deep right now. No one wants to use anything but nvda and cuda. Both AMD and Intel and many others have alternate HW and SW that can train and run all these models with reasonable performance and are far cheaper, but no one cares right now. NVDA forward PE is 27, which is extremely reasonable for the kind of growth people are projecting, so I wouldn't call it a bubble just yet. However, would I buy NVDA today? Probably not. Partly its because I already own it and made lots of money, but also when I buy individual stocks, I'm looking for potential for big gains and I don't see NVDA going from $3T to $30T.

 

[GrayHare]

IMO an important related question is whether human language operates in much the same way as the AI bots. Do we effectively have a large language model in our brains, trained by years of interaction with others? Is our biochemistry the primary difference between us and these bots?

 

[target2019]

I think we have roughly three groups of thought on this thread.
- It is hype and a bubble
- It is a stock market bump or bubble with specific companies that has potential to deflate, then eventually grow to incredible heights over a longer time period
- AI is here and now and we should go all in on anything AI

I don't really disagree with your post, except to say please give a little more credit to other institutions than just Stanford.

My mild skepticism on this thread is not about AI's ultimate potential, it is about the NVDA bubble right now. This chart alarmingly looks like CSCO or JNPR of year 2000. The talk is similar too. 25 years hence, both Cisco and Juniper are still alive, but both have not recovered their stock price even after a quarter century.

Yet, the internet has taken off and gone beyond what many thought possible. You see, CSCO and JNPR were not the only players in the game. Others came along, sometimes from surprising places. You know, that book club company called Amazon eventually sold something called AWS, and stomped on a market that CSCO should have invented. Most didn't see that coming.

My point: skepticism can come in the overall concept, which I think is a mistake. Skepticism can also come in investing in specific companies. That's me, and I suggest that care needs to be taken when investing in "AI". After all, don't forget the pets.com fiasco.

CSCO went through a few transformations along the way. In that age it was tech and promises. In this age it (along with other players) is necessary security suites which are the backbone of the internet.

People often use CSCO and how it never came back as an example of something or other. But the difference in mission and structure between then and now is what one should focus upon. It's no longer a fast grower as it once was. In 2000 it would have been in a growth fund, while today you'll find it in value, dividend and other focused funds.

AI is a software category that requires the hardware of other companies as it executes continuosly. As you say that will develop in leaps and bounds.

CSCO and all of its constituent results in a different category. Since the company is mature, it grows by acquiring other more-innovative companies. I call it an internet utility company. I'm not expecting its performance to exceed the 7 wonders. It's a different animal.

We are in a bubble, that's for sure. Bubbles happen in capitalism. And then they deflate. Transformative technology continues to evolve. It doesn't stand still.

 

[JoeWras]

IMO an important related question is whether human language operates in much the same way as the AI bots. Do we effectively have a large language model in our brains, trained by years of interaction with others? Is our biochemistry the primary difference between us and these bots?

You can be darn sure that if the AI entities ever declare war against us, we will not understand a dang thing they are communicating between each other. It won't be human language. It won't be "IPV6 protocol encrypted". It will be something we cannot even imagine.

 

[JoeWras]

Nvidia's moat is pretty deep right now. No one wants to use anything but nvda and cuda. Both AMD and Intel and many others have alternate HW and SW that can train and run all these models with reasonable performance and are far cheaper, but no one cares right now. NVDA forward PE is 27, which is extremely reasonable for the kind of growth people are projecting, so I wouldn't call it a bubble just yet. However, would I buy NVDA today? Probably not. Partly its because I already own it and made lots of money, but also when I buy individual stocks, I'm looking for potential for big gains and I don't see NVDA going from $3T to $30T.

I know I started the CSCO analogy, and I apologize for that. It isn't exact. Part of what happened to CSCO was that AMZN and GOOG developed their own chips and routers to take CSCO out of the equation. I can see that happening to NVDA. AI companies can steal talent from NVDA and elsewhere and develop in-house. It takes a few years, but not too many. They should learn a lesson from that.

 

[aja8888]

I know I started the CSCO analogy, and I apologize for that. It isn't exact. Part of what happened to CSCO was that AMZN and GOOG developed their own chips and routers to take CSCO out of the equation. I can see that happening to NVDA. AI companies can steal talent from NVDA and elsewhere and develop in-house. It takes a few years, but not too many. They should learn a lesson from that.

The Exec's know this.

This is the main reason the execs grant themselves stock options and take their money early, before the company becomes one on many in the field.