Edition #267 – June 4, 2023

Photo by Daniel Mitev

For some of us, tech provides an ever-evolving banquet table of opportunity. For others – it’s an ever-evolving invitation to destroy the fabric of our nation and our lives. 

Clearly, we are not educated to vet what is coming into our laps daily. 

Clearly, we are a gullible society influenced by high emotion coupled with information taken out of context and amplified.

Clearly, the above combination doesn’t bode well for making sane decisions. 

Stephen Hawking said: “Success in creating AI would be the biggest event in human history. Unfortunately, it might also be the last, unless we learn how to avoid the risks.”

Below is a Daily Pnut warning-report. 

By the way, I highly recommend Daily Pnut. I like the way the site structures the top news stories along with providing links for deeper dives into topics of interest. It is one of several newsletters published by Tim Hsia. Tim is a graduate of West Point, and was deployed twice to Iraq. After leaving active duty, Tim graduated from Stanford University’s JD/MBA program. He is also the co-founder of Service to School, a nonprofit for veteran higher education programs.

The Election Looms While Deepfakes Lurk

As the 2024 presidential election approaches, the waves of deepfake videos of presidential candidates and political figures are already beginning to wash ashore. We’re over a year out, but we’ve so far seen videos of Hillary Clinton, Joe Biden, and Anderson Cooper making insane statements (endorsing Ron DeSantis, ranting at a transgender person, and dropping profanities, respectively) using AI-powered fabrication technology called deepfaking. 

For those out of the loop, deepfaking uses AI to essentially create a 3D mask of, say, Joe Biden by studying the millions of images available of him online. It can then overlay this mask over the face of a different person in a video. Here’s an example of a deepfaked Tom Cruise – though it helps that the actor has his mannerisms down pat. There are also audio deepfakes, which use AI to replicate voices using recordings available online. 

DeepMedia, a company trying to develop tools to detect deepfakes, predicts there will be 500,000 video and audio deepfakes shared online in 2023 alone. Twitter, Facebook, and YouTube all have policies in place to take down harmful deepfakes, but it’s unclear if those will be enough to handle the presidential election – especially as deepfakes get easier to create. AI company OpenAI has already blocked its image generation tech from impersonating certain political figures, though competitor Midjourney has no such policy in place yet. On Tuesday, the Center for AI Safety – which includes leading members from the AI industry – released a (very brief) statement declaring that humanity needs to treat AI as a possible extinction-level event. “Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war,” reads the statement in its entirety. The statement adds to growing calls for increased regulation of AI, even as the technology seems to be in its early stages – deepfakes might just be the tip of the iceberg.

Related Posts