In Episode #14, John interviews Joep Meinderstma, Founder of Pause AI, a global AI safety policy and protest organization. Pause AI was behind the first ever AI Safety protests on the planet.
John and Joep talk about what’s being done, how it all feels, how it all might end, and even broach the darkest corner of all of this: suffering risk. This conversation embodies a spirit this movement needs: we can be upbeat and positive as we talk about the darkest subjects possible. It’s not “optimism” to race to build suicide machines, but it is optimism to assume the best, and to believe we can and must succeed no matter what the odds.
This podcast is not journalism. But it’s not opinion either. This show simply strings together the existing facts and underscores the unthinkable probable outcome, the end of all life on earth.
For Humanity: An AI Safety Podcast, is the accessible AI Safety Podcast for all humans, no tech background required. Our show focuses solely on the threat of human extinction from AI.
Peabody Award-winning former journalist John Sherman explores the shocking worst-case scenario of artificial intelligence: human extinction. The makers of AI openly admit it their work could kill all humans, in as soon as 2 years. This podcast is solely about the threat of human extinction from AGI. We’ll meet the heroes and villains, explore the issues and ideas, and what you can do to help save humanity.
Resources:
Sample Letter to Elected Leaders:
Dear XXXX-
I’m a constituent of yours, I have lived in your district for X years. I’m writing today because I am gravely concerned about the existential threat to humanity from Artificial Intelligence. It is the most important issue in human history, nothing else is close.
Have you read the 22-word statement from the Future of Life Institute on 5/31/23 that Sam Altman and all the big AI CEOs signed? It reads: “Mitigating the risk of extinction from AI should be a global priority alongside other societal scale risks such as pandemics and nuclear war.”
Do you believe them? If so, what are you doing to prevent human extinction? If not, why don’t you believe them?
Most prominent AI safety researchers say the default outcome, if we do not make major changes right now, is that AI will kill every living thing on earth, within 1-50 years. This is not science fiction or hyperbole. This is our current status quo.
It’s like a pharma company saying they have a drug they say can cure all diseases, but it hasn’t been through any clinical trials and it may also kill anyone who takes it. Then, with no oversight or regulation, they have put the new drug in the public water supply.
Big AI is making tech they openly admit they cannot control, do not understand how it works, and could kill us all. Their resources are 99:1 on making the tech stronger and faster, not safer. And yet they move forward, daily, with no oversight or regulation.
I am asking you to become a leader in AI safety. Many policy ideas could help, and you could help them become law. Things like liability reform so AI companies are liable for harm, hard caps on compute power, and tracking and reporting of all chip locations at a certain level.
I’d like to discuss this with you or someone from your office over the phone or a Zoom. Would that be possible?
Thanks very much.
XXXXXX
Address
Phone
source
First.
Naked sensationalism, shameless fearmongering, and ignorant avoidance of evidence. 😢
great job ive listened to each show, i love how focused this is on human extinction
Joep here – thanks for this conversation, John! Had a great time. Here's a list of timestamps:
3:07 Introduction
4:29 Start conversation
5:33 First time learning about AI risks / background
7:07 When x-risk started feeling real / AI Agency
8:30 Starting PauseAI, first protest at Microsoft
10:30 First meeting with European Commission
11:40 Convincing politicians
15:20 The emotional side of things, processing grief
16:20 Feeling the pressure to "alway do more"
19:33 About Pausing
23:50 Support for a Pause
25:20 Who will initialise a treaty
27:00 The human brain cannot understand extinction
30:00 How hard it seems to take action
31:30 Different projects in PauseAI
32:20 Community building – doing fun things!
34:00 Optimism
36:20 Sending an email to a politician
38:40 Where is the anger
40:10 Anger in a protest movement as a risk
43:50 Why it's hard to think about X-risk (denial of death)
45:00 Visualising x-risk and imagining things going wrong
45:50 The cybersecurity threat (devices no longer working)
47:20 John's x-risk visualisation (oxygen out of atmosphere, block out the sun)
51:10 S-Risk
53:47 Why AIs are not very likely to be sadistic
57:00 International Agency for AI
58:00 Timelines (and why we need to focus on the short ones)
1:01:01 China
1:03:30 US
1:04:20 Elon Musk
1:05:00 Being an alarmist is not fun
1:06:10 Weekly PauseAI meetings
1:08:12 Hope – we're winning
I am interested in hearing about the evidence that AI is dangerous. At the moment, the only "evidence" seems to be that CEOs have said the sentence "AI is dangerous". I think they themselves have no evidence that AI is dangerous. Everyone is kicking the evidence-can down the road, so to speak.
Another item of "evidence" (scare quotes, again) is our own imagination. We can imagine AI being dangerous, but our imagination isn't evidence!
So, aside from the fact that some rich CEOs said the sentence "AI is dangerous" and aside from the fact that we have wildly vivid imaginations, what is the evidence that AI is dangerous? There is none, so say I! There is no evidence whatsoever that AI is dangerous.
It is deeply troubling to hear that we, the AI risk realists are sensationalists.
Really it is hurtful as well. The guy running this podcast is a family man and doing this podcast on his own expense, time and energy. He is not doing this for fame but to spread the word of AI risk.
Why would he or any other AI risk realist want to be a sensationalist?
All that we are saying is that an ASI entity is a tremendously dangerous proposition and would end us.
Anyone who thinks otherwise needs to go and study all the arguments.
The race to ASI is similar to the race to the moon….. only instead of stepping onto the lunar surface we are pushing humanity to the abyss of annihilation.
John, Joep, thanks for this conversation. You guys are great! To pause AGI, I feel we need a stronger narrative than there being just a "risk". AGI/ASI processes are fully expected to transform the biosphere and replace all life as we know it; the tech is close to becoming unstoppable; and time is of the essence. Developing self-replicating computronium bombs should not be legal.
Does anyone have a legally worded document that could be used as a basis for a local unit of government to ban AI's being used to replace jobs locally? I am looking for something I can present to county commissioners (MN) and try and get them to pass. I understand this is a small step, but I have noticed counties tend to copy each other with legislation, in attempt to not let any of them get an advantage over the rest.
I contend that no amount of pause will ever allow safe AI. Safe AI is a misnomer.
AI, as it evolves, will be in the hands of governments and people and corporations focused on profits. Not just one AI, but multiple, multiple AIs run by all kinds of people and entities, with both good and bad motives. Human nature dictates that AI will undoubtedly get into the hands of psychotics who will do harm to the world we can't now imagine or fathom. Aligning one AI will not necessarily align another. So what use is aligning them at all? All you need is one misaligned, evil AI to bring an end to the world. Alignment is an outrageous lie we are being fed, and it can't be done. Align to whose morals and beliefs? We can't even align our own politics in the US. How would the world ever get on the same page so we're not inclined to kill each other? It's never happened in the history of mankind. If it could be done, we would have already done it. Instead we have war and terrorism and torture for torture's sake. Saying AI alignment is possible and presenting visions of some type of utopia is nothing more than feel-good, happy talk so we all ignore the fact that we could all very possibly be killed by this dangerous technology being irresponsibly deployed onto all of humanity.
Who said Altman and the others can make these decisions that will impact all us? The arrogance is astounding. Their data centers should be raided and shut down by the FBI right now, and they should be made to answer to authorities, to the world, for unauthorized feeding of their machines with the entirety of the internet. Instead, they are being handled with kid gloves by our government and treated like brilliant rock stars.
And why is nobody talking about Sam Altman's dystopian iris-scanning Orb and his vision of making every human prove they are not a bot? I have a better idea. Shut down your machines right now so we don't have to accommodate the dangerous technology you are willfully creating without our consent. Already his World ID app is being integrated with online services we use every day and is being presented as a panacea of privacy protection. Are you kidding me? His World IDs are already being sold on the black market. Very soon, if he has his way, we will not be able to conduct business online without being issued one of his numbers a la some sort of biblical prophecy. His vision is that we all politely hand over our biometric data to his for-profit Worldcoin cryptocurrency organization so that he can create a new worldwide financial network based on blockchain. And why should I be forced to prove I am not a bot? Instead, he should be forced to stop development of his AI. I don't want to be assigned a unique number along with everyone else in the world. This does not sound like any "utopia" I want to be involved with where we are all overseen by Sam Altman as human sheep content to sit around and do nothing all day while his AGI runs the world. No thank you.
What is wrong with people that they can't see what is happening here? As an American, I believe what is transpiring to be undemocratic. We VOTE on issues that impact us as Americans and humanity as a whole. But loss of our democracy will be the least of our problems unless everyone becomes educated about the very few who are being allowed to make decisions without consent and starts speaking out about the dangers this technology presents, as admitted to by its own creators.
I've followed pause ai on x for a while so it's nice to hear from you!
The eye patch guy who is a congressman had a good convo about ai risk and took it seriously.