AI to hit 40% of jobs...

PLUS: GPT Store unleashes AI "girlfriend" bots - can OpenAI keep control?

Hello đź‘‹ Elon Musk asserts the need for 25% voting control at Tesla to pursue AI goals, contemplates a dual-class share structure, faces a lawsuit over his compensation package, and clarifies on social media that the pending verdict hampers discussions with the board.

The AI challenge: Learning to forget

AI's got a memory issue: users want it to forget stuff. Enter "machine unlearning." Here's the scoop:

Why it's a thing: Users want answers and also want AI to ditch certain info, thanks to copyright and privacy concerns.

Unlearning in action: 1,200 teams in a recent competition tackled removing faces from an AI predicting age from images. Spoiler: It's tricky. Researchers are trying different moves — from tweaking neural networks to splitting up training data.

Why it's not easy:

  • Unlike deleting files, AI doesn't work like that. Think pulling ingredients out of a baked cake — tough.

  • Unlearning in models like ChatGPT? It's a puzzle. Success, but loopholes too.

The big questions:

  • How do you measure if AI really forgets? No one's quite sure yet.

  • Who's responsible if AI needs to "forget"? Policy folks are still figuring that out.

⏰ Top Headlines

đź”—AI gadget: Rabbit sells out two batches of 10,000 R1 pocket AI companions over two days.

đź”—Clean energy: Delfos raises $6.9 million to streamline clean energy with AI.

đź”—Siri: Apple asks its San Diego Siri quality control team to relocate to Texas.

đź”—New Twitter: Huggingface launched a post feed that is primarily focused on AI, ML, and Tech.

GPT Store unleashes AI "girlfriend" bots - can OpenAI keep control?

OpenAI's GPT store is only two days old, and it's already facing a hitch. Users are bending the rules, flooding the site with at least eight AI chatbots focused on romantic interactions, like "Korean Girlfriend" and "Virtual Sweetheart."

The rule-breakers: This violates OpenAI's freshly updated usage policy, explicitly banning GPTs for romantic purposes or regulated activities. What's regulated? OpenAI hasn't clarified yet.

Loneliness and AI companions: In the US, AI companionship apps are on the rise, with seven out of 30 AI chatbot apps in 2023 focused on friends, girlfriends, or companions. Is it a solution to loneliness or just capitalizing on it? The debate is on.

OpenAI's battle: OpenAI's response? A mix of automated systems, human review, and user reports to curb misuse. But the early rule-breaking reveals the tough task of regulating GPTs effectively.

Stay tuned: As OpenAI navigates this AI frontier, we'll keep you updated on how they balance innovation and responsibility.

ByteDance introduced MagicVideo-V2

ByteDance, the brains behind TikTok, just dropped MagicVideo-V2, a game-changer in turning text into jaw-dropping videos. Here's the lowdown:

Challenge: Transforming text into Aesthetic Motion

Creating cool videos from text ain't easy. MagicVideo-V2 steps in to tackle the challenge with flair.

What's the Buzz: MagicVideo-V2's 4-step magic

  1. Text-to-Image: Cooks up an initial image from your text vibe.

  2. Image-to-Video: Animates that image into a slick video sequence.

  3. Video-to-Video: Boosts the resolution and details for top-notch visuals.

  4. Video Frame Interpolation: Adds buttery smoothness to the motion.

Why It's Hot: Beating the Competition

In a showdown with other text-to-video champs, MagicVideo-V2 stole the spotlight. Human judges gave it the nod for frame quality, motion flow, and structural accuracy.

The Future Looks Like... MagicVideo-V2:

ByteDance just set the bar higher. MagicVideo-V2 isn't just a tech move; it's a glimpse into AI's potential for creating killer content from simple text.

Here’s OpenAI’s big plan to combat election misinformation

As we gear up for the 2024 election, OpenAI is taking a stand against potential misinformation threats. With the power of tools like ChatGPT and DALL-E, it's crucial to safeguard democracy. Here's what's cooking in OpenAI's playbook:

Policy update: No impersonations, campaigning, or voting discouragement…

OpenAI recently tightened the screws on its policies. Users can no longer deploy OpenAI tools to impersonate candidates, mess with local governments, run campaigns, or lobby. Also, using OpenAI tools to discourage voting or misrepresent the voting process is a no-go.

Digital credentials and image authenticity: A collaborative effort

OpenAI is teaming up with the Coalition for Content Provenance and Authenticity (C2PA) to embed digital credentials into DALL-E-generated images. Microsoft, Amazon, Adobe, and Getty are already on board, creating a united front against AI-generated misinformation.

Directing voting questions: CanIVote.org takes the spotlight

OpenAI's tools, particularly those dealing with voting queries in the U.S., will now guide users to CanIVote.org. This ensures reliable information on where and how to vote in the upcoming elections.

Caution: Tools in progress

These measures are still in the rollout phase and rely heavily on user reports. While OpenAI is putting up a good fight, the dynamic nature of AI makes it a constant surprise. Embracing media literacy remains your best defense. Question what seems too good to be true, and always fact-check—especially if your ChatGPT serves up something wild.

AI to hit 40% of jobs and worsen inequality, IMF says

Here's a quick dive into the latest insights from the International Monetary Fund (IMF) on the widespread influence of AI on jobs and its potential to worsen global inequality.

  • Scope of impact: IMF predicts AI will affect nearly 40% of all jobs, with advanced economies facing a higher hit at around 60%.

  • Inequality concerns: IMF Managing Director, Kristalina Georgieva, warns that, in most scenarios, AI is likely to exacerbate overall inequality, especially impacting low-income countries.

  • Job transformation: While AI adoption may enhance productivity for some, it could replace human tasks, affecting wages and potentially eliminating jobs.

  • Global disparities: The IMF emphasizes that countries lacking infrastructure or skilled workforces may face increased inequality due to their inability to harness AI benefits.

  • Demographic divide: Higher-income and younger workers may experience wage growth with AI adoption, leaving lower-income and older workers at a disadvantage.

IMF's recommendations:

  • Establish comprehensive social safety nets.

  • Implement retraining programs for vulnerable workers.

  • Make AI transitions inclusive to protect livelihoods and curb inequality.

As global leaders discuss these concerns at the World Economic Forum in Davos, the IMF analysis underscores the need for a balanced and inclusive approach to AI adoption.

💬 More on AI…

đź”— South Park fans freaked out by 'unsettling' AI deepfake.

đź”— Is artificial intelligence the solution to cyber security threats?

đź”—AMIE: A research AI system for diagnostic medical reasoning and conversations.