I’ve been running a weekly meetup for our local rural community to discuss the changes happening from AI. We talk about the impact on jobs, food security, and more. The changes are not theoretical; they are already happening.
The community has responded well - awareness of risks has grown, along with resiliency planning.
Following requests from colleagues and online friends, I have started an online version of the meetup. Once a month I am hosting an opportunity to discuss your concerns and learn about the changes happening due to AI.
Bring your questions, bring your skepticism - and join the conversation.
The next meetup is Tue Jan 20, 6pm EST (Jan 21, 10am AEDT).
What concerns you most about the changes? Are you seeing shifts in your industry already? Your perspective matters, and others in your situation need to hear it.
This isn’t speculation or fear-mongering. Through my work developing AI systems and direct conversations with researchers at leading AI labs, I’m sharing what I’m actually seeing happen - information that’s public but not widely discussed outside tech circles.
Thanks for posting this, Charlie. I noticed that this will take place on Google Meet. Would that require me to install any Google apps on my Mac, or can I attend from a web browser like Safari or Firefox?
Just an FYI on Zoom… on at least some of their service versions, it is possible to join by browser. They just don’t make it easy to see how. I think clicking on the link that suggests you want to join by app will actually offer a way to do so using a browser as an alternative.
Tip: occasional or casual users of Zoom do not have to install the Zoom client. Zoom will run in most web browsers with no downloads or installations needed.
Simply follow the conference link sent out by the meeting’s host. Then wait until a “Join in browser” link appears in the browser window. If Zoom automatically downloads a software installer, avoid clicking on the installer and move it to the Trash.
Sometimes the “Join in browser” link doesn’t appear automatically. If so, try clicking on the Join Meeting button (yes, even though you haven’t installed the Zoom software).
The only requirement is that the host needs to have the join-in-browser option enabled. In my experience, very few hosts turn the option off.
This video is a good introduction to some of the topics we discuss, in particular AI alignment and challenges with pausing development of more advanced models. If you find the topics interesting, join us and share your thoughts and questions.
@cngarrison
You and your group might find this webinar interesting as the basis for a future discussion. It is about how AI algorithms and models can end up behaving more like subjective humans than objective robots, which has all sorts of effects in activities such as screening job applicants and making decisions.
The issue of model bias is directly related to human alignment, which is the critical points we discuss in the meetups. The importance of alignment goes further than bias though; I consider it to be the most critical discussion. Undesirable alignment, combined with arrival of AGI (singularity) could rapidly bring about a dystopian existence.
For anyone interested, the next meetup is Feb 18, although timezone difference will be difficult for USA. The March event will be better timezone for USA.
This thread is the first time I’ve seen that specific term used to describe the challenge of ensuring AI’s, including generative AI’s, do not do undesired things. So, I had a look around online. For anybody interested, here are two non-influencer/non-pundit articles I liked:
And Bruce Schneier, the cryptography and security expert, has been thinking about the impact of AI on privacy, security, and political systems for quite some time. For example:
Thanks for sharing your learnings on the topic. It is critically important that people have greater awareness around the risks (& potential benefits) of AI and alignment.
To be clear, I work with AI daily, building an AI product - I am not against AI advancements. But my awareness of those advancements is why I feel so strongly about having the discussion.
For anyone wanting to dig deeper into why AI alignment is a nearly insurmountable problem, I highly recommend the book “The Alignment Problem”. The issue was recognised last century, and we still don’t have reliable or guaranteed solutions. Which is why “AI security” is critical. It is also why some of the original OpenAI founders left to create Anthropic - they don’t consider AI alignment to be an optional afterthought.
The Anthropic CEO recently wrote another essay regarding AI risks, that go beyond alignment (but alignment is still a fundamental element).
Repeating myself from earlier post… if you have questions, opinions or skepticism about AI impact on our future, join the discussion. The online meetup is monthly (time alternates between morning and evening each month).
They aren’t recorded, simply because it’s a discussion format rather than presentation.
I’m hosting them monthly in hopes that people can make the next one if time doesn’t suit.
Alternatively, please suggest another day/time that is suitable and help organise people to attend.
My objective is shared awareness, and I’m flexible how that is achieved.
I have also thought a website to support the discussions, which would be suitable for people who simply don’t want to join a live discussion; but that is a non-trivial task which I haven’t yet made time for. It is a priority though.
I suggested they can join without camera.
