Two Minute Papers: The video discusses a new AI research paper on generating future scenarios using AI models, which can be run at home for free and is useful for training AI in self-driving cars and robots.
Jeff Su: LinkedIn and Austin Bellac released a report with 93,000 data points on job offer negotiation, highlighting the importance and techniques of salary negotiation.
Two Minute Papers - NVIDIA Cosmos - A Video AI…For Free!
The AI system described in the video can generate future scenarios from input images or text prompts, creating videos that help train AI systems like self-driving cars and robots. This is crucial for handling rare scenarios that lack sufficient real-world video data. The system is open-source, allowing users to run it at home for free, even for commercial purposes. It is designed to be easily fine-tuned for different hardware and use cases. Despite its potential, the system has limitations, such as slow generation times and imperfect video quality, but it represents a significant step forward in AI research. The paper detailing this research is available for free and includes user study results showing favorable comparisons to previous techniques.
Key Points:
- AI system generates future scenarios from images or text, aiding AI training.
- Open-source and free to use, even commercially, allowing home use.
- Helps solve rare scenario training for self-driving cars and robots.
- System is easily fine-tuned for different hardware and use cases.
- Current limitations include slow generation and imperfect video quality.
Details:
1. 🔍 Unveiling AI's Future Potential
1.1. Introduction to AI Research Paper
1.2. AI System with Multiple Models
1.3. Image to Video Transformation
1.4. Text2World Results
1.5. Output Quality
2. 🚗 Revolutionizing Robotics and Self-Driving Cars
- The system is open and accessible, allowing users to run it at home for free, promoting widespread usability and experimentation.
- Unique results can be generated using this technique, offering insights not available elsewhere.
- Although the visual quality may not match OpenAI’s Sora, the system is optimized for a different purpose, highlighting its effectiveness in specialized applications.
3. 📹 Generating AI Training Scenarios
- AI systems, such as self-driving cars and robots, encounter a long-tail problem characterized by insufficient training data for rare scenarios.
- A notable example includes AI misinterpreting a moving traffic light on a truck, highlighting the need for specific training videos to address such anomalies.
- These challenges arise because AI lacks the intuitive understanding humans possess, necessitating targeted training to improve AI's comprehension of uncommon situations.
- To enhance AI performance, it's crucial to create and integrate training scenarios that cover a broader spectrum of rare events AI might encounter in real-world applications.
4. 💻 Open Source AI: Accessible and Customizable
4.1. AI Training with Diverse Data
4.2. Realism in AI-Generated Content
4.3. Open Source AI Model Availability
5. 📜 Understanding AI's Boundaries and Rules
- The AI system's open-source nature allows for easy fine-tuning and development of custom variants, enabling adaptations for specific use-cases.
- The freely accessible research paper provides crucial insights into the system's development and capabilities.
- Understanding the limitations discussed in the research paper is essential to grasp the system's constraints and potential applications.
- Customization can be applied across different industries, enhancing product development cycles and operational efficiency.
- Open-source customization can face challenges like ensuring security and compatibility across different platforms.
6. 🔧 Overcoming AI Simulation Challenges
- AI models for simulation have manageable sizes, between 7-14 billion parameters, allowing them to run on high-end laptops.
- Despite manageable sizes, generation times are slow; a consumer graphics card may take 5 minutes to produce a few seconds of video.
- The quality of AI-generated results is currently low, with issues such as incorrect physics (e.g., floating objects, extra fingers) and lack of object permanence.
- An autoregressive technique offers faster generation but compromises visual quality.
- There is significant room for improvement, emphasizing that research is an ongoing process.
7. 🎉 Celebrating AI Advancements and Future Directions
7.1. AI Speed and Accuracy Improvement
7.2. User Study and Community Contribution
7.3. Implications and Future Directions
Jeff Su - Negotiate a Higher Salary: 3 Evidence-Based Tips!
The report emphasizes the significant financial benefits of negotiating job offers, revealing that individuals who negotiate can earn up to twice as much over their careers compared to those who don't. Despite this, 54% of candidates do not negotiate their salaries. The report provides evidence that 84% of employers expect negotiations and have budgeted for them, with 90% never withdrawing an offer due to negotiation attempts. Practical negotiation techniques include avoiding early salary discussions, preparing a bolstering range using data from platforms like Glassdoor and H1Bdata, and adhering to the 'double nope' rule, which involves being prepared for three rounds of negotiation. The report also highlights that 93% of candidates who negotiate receive more than the original offer, underscoring the importance of negotiation in maximizing compensation.
Key Points:
- Negotiating can double career earnings, yet 54% of candidates don't negotiate.
- 84% of employers expect negotiations; 90% never withdraw offers due to negotiation.
- 93% of candidates who negotiate receive more than the original offer.
- Avoid early salary discussions and prepare a bolstering range for negotiations.
- Use the 'double nope' rule: be ready for three rounds of negotiation.
Details:
1. 📊 LinkedIn's Negotiation Report Overview
1.1. Report Creation and Data Points
1.2. Focus on Practical Techniques
1.3. Objective of the Report
2. 💼 Takeaway 1: Importance of Negotiation
- Individuals who negotiate their salaries can earn up to 100% more over their careers compared to those who do not negotiate.
- 54% of candidates do not negotiate their salary, resulting in significantly lower lifetime earnings.
- The average salary in the US is approximately $63,000, and negotiating can yield a 10% to 20% pay increase when changing jobs.
- If two individuals start with a $63,000 salary, and one negotiates a 10% increase every time they change roles every 5 years over a 40-year career, the difference in cumulative earnings can be as much as $1.9 million.
- By year 40, the annual salary for a person who negotiates can be $154,000 more than someone who does not.
3. 🤔 Takeaway 2: Overcoming Negotiation Fears
- 84% of employers expect candidates to negotiate and have budgeted for it.
- The top three reasons people don't negotiate are lack of confidence (33%), fear of seeming greedy (35%), and fear of losing the offer (50%).
- 90% of employers have never withdrawn an offer just because someone negotiated.
- 73% of employers are not offended by negotiations.
- The company's initial offer should be viewed as an opening bid, not the final one, indicating openness to salary discussions.
4. 📈 Takeaway 3: Success Rate of Negotiation
- 93% of candidates who negotiated received offers higher than the original offer.
- 46% of candidates received offers exceeding the expected $70,000 after negotiation.
- 47% of candidates received offers less than $70,000 but more than the original $60,000 offer.
- Negotiation often results in a higher salary than initially offered, highlighting its importance for job seekers.
5. 🛑 Tip 1: Avoid Early Salary Discussions
- Companies often ask for salary expectations early to disqualify candidates and save resources. Candidates should tactfully delay these discussions to maintain leverage.
- By asking for the budgeted salary range for the role instead of disclosing their own expectations, candidates can defer the conversation.
- Recruiters usually share salary ranges to avoid mismatches, making it a viable strategy to inquire about it upfront.
- Pretending to negotiate for a friend can help candidates feel more comfortable during salary discussions.
- To delay salary discussions, candidates can use phrases like 'I'm more interested in understanding the role and the team first.' or 'Can you share the budgeted range for this position?'
6. 🔍 Tip 2: Prepare Your Salary Range
6.1. Research Salary Ranges
6.2. Application of Salary Data
7. 🔄 Tip 3: Use the Double Nope Rule
- Implement a structured negotiation approach, anticipating three rounds of responses to an offer.
- Start with Plan A, your ideal compensation package, and be prepared to negotiate if declined.
- Transition to Plan B by emphasizing increased equity and bonuses if initial requests are not met.
- Finally, move to Plan C, focusing on non-monetary benefits such as PTO, remote work flexibility, and education budgets if previous plans are rejected.
- Use a negotiation script to maintain professionalism: 'I understand the constraint around base and bonus. I want to be respectful of your budget, but I'd like to make sure my compensation is aligned with my value.'
8. 💡 Bonus Tip: Seek Recent Experiences
- Take advice from people who are just one or two steps ahead of you for actionable insights.
- The best negotiation advice comes from those who recently experienced the same process.
- Reach out to professionals on LinkedIn in your target role for guidance.
- Watch videos on how to connect with professionals without appearing pushy.
