Image produced by Myshoun, using AI. Found on Pixabay.
We are at an interesting but predictable point in time with AI. Are we on the launch ramp or on the edge of the cliff? Yes.
What We Heard and Saw in Recent Months
It seems like only yesterday that Artificial Intelligence (AI) popped up into popular consciousness. In fact, it has been around for a few years, with remarkable growth in popularity, utility, and dangers. As Bill Gates has said, everything you hear about the promise and potential of AI is true; also, everything you hear about the dangers of AI is true.
On a scale perhaps equal to or beyond atomic energy, the power and risk of AI for humanity is at the extraordinary end of possibilities. What AI is already doing in medicine and science is breath taking. So too are its foibles in bias, error rates, and what feels like absolute confidence in its own answers.
Then Two Big Changes: China and GPT-5
China surprised everyone just a few months ago with its launch of its premier AI chatbot system, DeepSeek. As near as we can tell, they developed in record time and managed to completely work around US restrictions on AI technology and hardware. Deep seek is fast, highly capable and costs much less than its competitors.
Since its introduction, China has continued upgrades and has introduced several more AI systems, with more on the way. It is astonishing to me that people have been quick to adopt this tool, with little concern about what information China obtains when you use their system. In any case, this is another example of the kind of explosive growth and change that is built into the AI world.
Look at something of an opposite effect in GPT-5, the latest AI chatbot update from Open AI. GPT-5, and its predecessor, is what started the revolution in popular use of AI. People have been waiting breathlessly for version 5, expecting yet another quantum leap in capability. The expectation generally in AI has for some time been that explosive, quantitative growth in sophistication was to be the norm, with every iteration representing a previously unimaginable breakthrough. But this has proven not to be the case.
GPT-5 is certainly an improvement, but only incrementally. People who use AI a lot have been underwhelmed if not disappointed. The case made years ago for the Singularity, wherein AI becomes sentient and people/computers merge, laid out a model with compound growth at every stage. But this latest round shows this is not always so.
This Pattern Was Called Out a Long Time Ago
In fact, we should have foreseen the occasional one step forward, one step back process now in evidence. Alvin Toffler, in his remarkably prescient and accurate view on progress in his book Future Shock pointed out that new technology almost always follows a predictable pattern.
First, something new comes out that seems like Sci-Fi. Only a few believe in the new technology, most see it as unreal. Then the new technology has a breakthrough, and everyone gets excited and on board. Then the new technology slows down and has set backs and many give up on it. Lastly, the new technology works out the bugs and succeeds beyond almost everyone’s imagination.
My read is that AI is in that next to the last stage of a seeming roadblock – it will not last long. Startling progress is coming, in so many areas. And all the risk that comes with it also awaits us.
So, Is This Stuff Working or Not? What is the Future?
Yes, the Singularity is still likely in our future, and not that far off. This stuff really works. The potential in almost every sphere is mind boggling. Already AI is doing somethings that its creators are not sure how it makes such things happen. We see glimpses already in science, research, medicine, the arts, crime, government, social interactions, and more. I have been at least playing with several of these outlets in a couple of spheres and I see real value and potential.
But I also know that such systems have built in biases from their human creators (maybe when AI is more developed by AI, this problem will reduce). I know that on average, something like 20% of AI content could be erroneous. And I know the old rule of “garbage in, garbage out” applies here, too. How well you pose a question has a major effect on the quality of answers returned to you. Buyer beware.
The impact on global society is huge. In earlier epochs, automation threatened and eliminated lots of jobs, mostly in industry and manufacturing, but in other areas as well (talked to a telephone operator or 411 persons lately?). Next, it looked like truck drivers and taxi drivers were about to be replaced. In due course, they will be replaced.
But now we also find artists, actors, accountants, lawyers, and others looking at going extinct. It is likely none of us going forward will trust any “photos” we see or “recordings” we here. People are already creating amazingly human like AI friends or partners, with all that entails.
We have a lot to figure out. As usual, our technology is advancing faster than our understanding of how to use it wisely. I expect that we will get there, eventually and unevenly, but we likely will get there before AI decides it no longer needs humanity.
The potential is worth the risk. In any case, there is no putting the genie back in the bottle. Those who would an AI from education or the arts, for example, will fail. Get over it and figure out how to work and live with this new partner to humanity. It is not going away and it will not be very much contained. Ride it or get trampled by it. I prefer the first choice.
Bill Clontz
If you find this blog worthy of your time and curiosity, I invite you to do three things:
(1) Join the conversation. Your voice counts here. If you wish to share COMMENTS anonymously, make the last word in your comment “PRIVATE.” I will assure your privacy via anonymity.
(2) Share the word about this post with friends and colleagues. Share a link in your emails and social media posts (https://agentsofreason.com).
(3) You are welcome to share this post with anyone. It is easy to pass on via email, of course, but also on Facebook Blue Sky, LinkedIn, or Reddit; simply click on the links for these services at the end of this article.
Let’s grow our circle.
You gave a very succinct description of the issue. I am currently using AI in my new music journey. It absolutely can be useful, amazing, wonderful. It is an invaluable tool for me and made the entry barriers to music as an older adult with zero previous experience in either music or AI tech much easier to overcome. I also heard an interview with director/writer James Gunn in which he gave some insight to the technology as it pertains to the industry of movies, television, streaming and all other aspects of that industry. I agree it is either a tool we will wield or be ruled by, dealers choice. I am personally excited by the opportunities it presents. As far as being a source of information, the people who already question rather than consume will be just fine. It is the rest of the populations willingness to be given rather than seek information and confirmation that may cause a problem. As my father, Pete, is fond of saying, there were people calling the invention of the car the end of civilization as we know it not so long ago.