That "the more powerful, the more important it is that everyone has access"?
Especially considering that the biggest killer app for AI could very well be smart weapons like we've never seen before.
Nukes aren’t even close to being commodities, cannot be targeted at a class of people (or a single person), and have a minutely small number of users. (Don’t argue semantics with “class of people” when you know what I mean, btw)
On the other hand, tech like this can easily become as common as photoshop, can cause harm to a class of people, and be deployed on a whim by an untrained army of malevolent individuals or groups.
It's important to weigh the benefits of diversity and open competition against the risks of bad actors misusing the tools. Ultimately, finding a balance between accessibility and responsible use is key.
What guarantee do we have that OpenAI won't become an evil actor like Skynet?
The direct threat to society is actually this kind of secrecy.
If ordinary people don't have access to the technology they don't really know what it can do, so they can't develop a good sense of what could now be fake that only a couple of years ago must have been real.
Imagine if image editing technology (Photoshop etc) had been restricted to nation states and large powerful corporations. The general public would be so easy to fool with mere photographs - and of course more openly nefarious groups would have found ways to use it anyway. Instead everybody now knows how easily we can edit an image and if we see a shot of Mr Trump apparently sharing a loving embrace with Mr Putin we can make the correct judgement regarding a probable origin.
I was replying to a comment saying that nukes aren't commodities and can't target specific classes of people, and I don't understand why those properties in particular mean access to nukes should be kept secret and controlled.
What do you mean? Are you being dramatic or do you actually believe that the US government will/can not absolutely shut OpenAI down, if they feel it was required to guarantee state order?
...after amazing public world wide demos that show how real the AI generated videos can be? How long has Hollywood had similar "fictional videos" powers?
if it were opened to public faking such videos would lose (nearly) all of its power
How quickly do you think our gerontocracy will adapt to the new reality?
When a significant fraction of video is generated content spat out by a bored teenager on 4chan, then people will stop trusting it, and hence it will no longer have the power to convince people to kill.
Isreal media machinery parading photographs of damaged houses that could only be done by heavy artillery or tank shells blaming on rebels carrying infantry rifles.
But I agree, as if the current tools were not enough to sway people they will have more means to sway public opinion.
Not directly. But I won't be surprised if AI video generators aren't somewhere in the chain of causes of gigadeaths this century.
A homing missile that chases you across continents and shows you disturbing deepfakes of yourself until you lose your mind and ask it to kill you. At that point it switches to encourage mode, rebuilds your ego, and becomes your lifelong friend.
The ideas of critical mass, prompt fission, and uranium purification, along with the design of the simplest nuclear weapon possible has been out in the public domain for a long time.
As things are at the moment, while supression of a technology has benefits, it seems like a risky long-term solution. All it takes is for a single world-altering technology to slip through the cracks, and a bad actor could then forever change the world with it.