Beginners in AI

Thank you for joining us again!

Welcome to this week's edition of Beginners in AI, where we explore the latest trends, tools, and news in the world of AI and the tech that surrounds it. Like all editions, this is human curated, edited, and published with the intention of making AI news and technology more accessible to everyone.

We have a top story on Google’s updated approach to AI weapons, plus coverage on a million-dollar challenge from Perplexity, Tesla’s Dojo progress, Anthropic’s jailbreaking tests, Elon Musk’s AI-Doge curiosity, a medical procedure bridging Rome and Beijing, and a new Figure for the company gunning for Tesla’s Optimus.

AI TOP STORY 
Google Drops AI Weapons Ban: What It Implies for the Future of Artificial Intelligence and Warfare

Background
Google previously set guidelines for AI usage in its AI Principles, avoiding involvement in weaponry among other things. This approach was praised by many who saw it as a step toward responsible AI development and derided by those believing they were depriving the United States of valuable technology should war with China ever occur.

What Changed
The update suggests Google may be more open to government contracts linked to military or security projects. The attention regarding this announcement is second only to the time that Google removed the “don't be evil” preface form its official code of conduct in 2018. Google is not alone in this shift. OpenAI, for example brought on former National Security Agency chief Paul Nakasone to its board and hired former Defense Department official Sasha Bakerhead in a move that was speculated to make it easier for them to procure government defense contracts. In a move that may or may not be coincidence, China has opened an anti-monopoly investigation into Google only five days ago.

Why to Take Away
Some believe this indicates a broader trend of tech giants exploring new ways to bring in AI revenue without resorting to more ads. It's also possible that Google, with their bird's eye view of billions of searches throughout the internet, know something that the rest of us do not. In a recent interview, Former Google CEO Eric Schmidt described the future of warfare as drone on drone. Elon Musk shared similar sentiments in an interview he did at West Point earlier this month. Regardless of the hardware in question, it will be AI technology at the helm.

LAST WEEK IN AI AND TECH 

Perplexity’s Million-Dollar SuperBowl Play
Perplexity, the AI-powered search engine known for its question-answering model, is offering a jaw-dropping prize for asking questions. Recently in the news for offering to buy TikTok, CEO Aravind Srinivas made the announcement on X that the company would offer a million dollar drawing instead of taking out a traditional Super Bowl ad. Starting at 3 p.m. on Super Bowl Sunday(today), participants get a shot at $1 million by posing questions on their mobile app. Each person can receive up to 26 entries. Sweepstakes entries are calculated as follows: One entry for downloading the app, creating an account, and entering the sweepstakes; one entry per app referral (capped at 20); and five extra raffle entries for asking five questions during the game. Eligibility for the Perplexity Big Game: You must be in the United States. No need to be a Pro user. You must have a signed-in account with push notifications enabled.

Read More

Dojo’s Code Escape
Tesla's Dojo project aims to build a supercomputer for training full self-driving systems, designed to process vast amounts of video data from Tesla's vehicle fleet. Elon Musk envisions a future where Tesla's vehicles learn and adapt on the go, potentially leading to more robust performance and fewer errors in autonomous driving. "Dojo is optimized for machine learning, particularly the video neural networks that are critical to interpreting camera inputs in Tesla's self-driving system". Dojo is big part of Musk's plan for Tesla’s Robotaxi fleet and maybe even his Optimus robots that the company has been gearing up for( Rise of the Machines: Robotaxis and Humanoid Helpers)

Read More

Unlocking Claude: The Great Jailbreak
Anthropic, the creator of Claude AI, is inviting users to uncover possible flaws in its language model's guardrails. The project's aim is to test how easily the AI can be manipulated into breaching guidelines. Anthropic's CEO notably came under withering criticism from Marc Andreessen recently, who said, "Fear mongering for regulatory capture and to kneecap open source AI. The existential threat of open source AI is to the Big AI cartel." Andreessen was referencing a video clip of Anthropic CEO Dario Amodei claiming that the open-source model DeepSeek that took the internet by storm last week, generated "potentially dangerous information." Anthropic has emphasized that their mission is safety above all else. It is one of the reasons cited for Claude not being able to access the internet on the native platform, when competitors ChatGPT and Perplexity can.

Read More

Musk + Doge = AI Meme?
Elon Musk's Department of Government Efficiency (DOGE) is making waves in the federal government by implementing AI technology to analyze government spending. The initiative involves training a custom AI chatbot for the US General Services Administration to sift through contracts and procurement data. Machine learning is already used to detect patterns in everything from movie choices on Netflix to fraud detection in financial transactions, medical diagnosis in healthcare, and autonomous driving in transportation. As Thomas Shedd, head of the government's Technology Transformation Services, stated, "This goes back to this, how do we understand how the government is spending money?"

Read More

Surgery from Afar
A medical team in China led by Zhang Xu carried out a urology operation remotely, connecting surgeons in Beijing to a patient in Rome through specialized robotic tools. High-speed networks combining 5G and fiber optics helped them make precise movements with imperceptible lag-time, even though they were an estimated 5000 thousand of miles apart. This event demonstrates how health care can expand across borders and possibly save lives in remote locations or countries short on skilled surgeons. It’s a glimpse into a future where cutting-edge networks and robotics join forces to provide life-changing procedures.

Read More

TECH TERMS TO KNOW

Jailbreaking is when someone removes a device or AI model’s restrictions so it can do tasks it wasn’t originally meant to do. On smartphones, it might mean installing apps that aren’t approved by the phone’s official store. With AI, “jailbreaking” can allow hidden prompts or responses.

Real-World Examples

iPhone Jailbreaking: In 2007, just months after the first iPhone's release, hackers developed methods to jailbreak the device, allowing users to install third-party apps before the official App Store existed.

PlayStation 3 Jailbreak: In 2010, a group of hackers successfully jailbroke the PlayStation 3 console, enabling users to run unauthorized software and pirated games.

GPT-3 Jailbreaking: Researchers discovered ways to bypass GPT-3's content filters by using carefully crafted prompts, allowing the AI to generate content it was designed to avoid, such as explicit or biased material.

TOOL SPOTLIGHT (non-sponsored)

Venice AI is a private and uncensored alternative to popular AI applications, offering a range of AI-powered services while prioritizing user privacy and freedom of expression. 

What is Venice AI?

• Text generation and editing

• Document summarization

• Research assistance

• Image creation

• Code generation and debugging

How it works

Venice AI operates on the following principles:

1. Privacy-focused: All user prompts and data are stored locally on the user's device, never on Venice's servers.

2. Uncensored AI: The platform utilizes leading open-source AI technology to deliver unbiased machine intelligence without censoring responses.

3. Versatile tools: Users can access multiple AI models for various tasks, including text and image generation.

ROBOTICS AND AI

Figure AI’s Big Order
Figure AI has announced plans to scale up robot production to 100,000 in the next four years to meet the demand for a new customer that CEO Brett Adcock claims is “one of the biggest U.S. companies.” The company's Figure 02 model has been out since last year and figure 03 was confirmed production-ready back in October. Figure’s model was covered in last week's newsletter Copyright Office: AI Content is Protected, If Humans Helped comparing it to its main competitors in a 2024 robot visual lineup and comp table.

TRY THIS PROMPT (copy and paste into AI ChatGPT, Grok, Claude, Gemini)

Problem Solving:

I'm facing a challenge at work/school where [describe the challenge]. 

First, brainstorm three different approaches I could take to solve this problem, along with the potential pros and cons of each. 

Consider factors such as time constraints, resource limitations, and potential risks. 

Then, suggest alternative solutions that might be less conventional but could potentially lead to better outcomes. 

After that, provide a step-by-step plan for implementing the chosen solution, including specific actions and timelines. 

Lastly: 
1.provide a contingency plan in case the chosen solution does not work as expected. Consider also the ethical implications of each approach and suggest ways to ensure that the chosen solution is aligned with ethical principles. 
2.provide a list of resources that could be helpful in implementing the chosen solution. This list could include books, articles, websites, or people who could provide guidance or support." (Practical for problem-solving)

DID YOU KNOW?

The first website is still online. Sir Tim Berners-Lee, the inventor of the World Wide Web, created the world’s first website in 1991. It's available to view today at info.cern.ch.

30 Referrals: Lifetime access to all Beginners in AI videos and courses

AI-ASSISTED IMAGE OF THE WEEK

Interested in stock trading, but no clue where to start? Check out our sister newsletter launching daily in March.

Thank you for reading. We’re all beginners in something. With that in mind, your questions and feedback are always welcome and I read every single email!

-James

By the way, this is the link if you liked the content and want to share with a friend.