Chinese chatbot thinks it’s ChatGPT!

Hi pals! 🎄🎅 Wishing all of you a Merry Christmas in advance! May your holidays be filled with love, laughter, and warmth. Here’s what you need to know about AI in these last remaining days of 2024:

👉 Chinese chatbot thinks it’s ChatGPT

👉 Google doesn’t want to lose to OpenAI

👉 Nvidia wants to build humanoid robots

📧 Did someone forward you this email? Subscribe here for free to get the latest AI news everyday!

Read time: 5 minutes

What’s going on: In a recent BBC interview, Geoffrey Hinton, known as the "Godfather of AI," is warning that the rapid growth of artificial intelligence could be dangerous for humanity. He believes that there is now a 10% to 20% chance that AI could cause human extinction within the next thirty years. Hinton, who won a Nobel Prize for his work in AI, thinks we need strict rules to control AI development because companies are mainly focused on making money, not on safety.

What does it mean: Hinton emphasizes that we've never had to deal with things more intelligent than ourselves before. In his view, just leaving AI to the profit motive of large tech companies is not going to be sufficient to make sure they develop it safely. (This actually might be very true, as we learned last week that OpenAI and Microsoft secretly define AGI as a system that’ll be able to make them billions of profit).

More details: There are already over 120 proposed laws in the US aimed at regulating AI technology. The Biden administration has made some efforts to manage AI safely, but there are many challenges remained. For example, a recent law in the European Union was criticized for having too many loopholes. Hinton emphasizes that only government regulation can ensure AI is developed responsibly.

What’s going on: Last week, a Chinese AI lab, DeepSeek, introduced a new open-source AI model called DeepSeek V3, which beats many other open-source models in terms of size, accuracy, speed, and lower costs of training. In tests conducted by TechCrunch and posts on X, the chatbot insists it is a version of OpenAI's GPT-4 released in 2023! If you ask about DeepSeek’s API, it offers instructions for using OpenAI’s API instead. It even shares jokes that are identical to those from GPT-4.

What does it mean: Both ChatGPT and DeepSeek V3 are statistical systems trained on vast datasets. While DeepSeek has not disclosed the specifics of its training data, it’s possible that the model has absorbed outputs from public datasets containing text generated by GPT-4. Moreover, OpenAI's terms of service prohibit using its outputs to develop competing models.

More details: 

• DeepSeek V3 boasts an impressive 671 billion parameters.

• It was trained in just two months at a cost of $5.5 million. Significantly less than its competitors.

• DeepSeek trained V3 on 11.1 trillion words with the help of Nvidia H800 GPUs.

• It has shown remarkable performance in coding competitions, outperforming models like Meta’s Llama 3.1 and OpenAI’s GPT-4o.

• The web is increasingly filled with AI-generated content, complicating the training newer AI system like DeepSeek V3 which might be biased by being trained on the outputs generated by GPT models.

What’s going on: Google's CEO, Sundar Pichai, is sounding the alarm bells. At a recent meeting with Google employees, he emphasized the critical importance of 2025 for the company's AI future. Pichai fears that ChatGPT is quickly becoming the word for AI, much like "Google" has become synonymous with internet search. This poses a serious challenge to Google's dominance in the tech world.

What does it mean: AI race is getting tense, and Google doesn’t want to fall behind and lose its industry leader status to other tech companies, most importantly OpenAI. Google was actually the first company that provided the world the key technical enabler of current AI systems, a neural network architecture called “Transformers“, and also the required data for training next generations of AI systems. Therefore, it is only fair for Google to want to maintain its dominance in the world of AI.

More details: 

• Pichai stressed that scaling Gemini for consumer use will be their top priority in 2025.

• They might integrate Gemini into Google Search, creating a separate AI-powered search mode, or developing entirely new consumer-facing AI applications.

• Pichai acknowledged that the race for AI supremacy is dynamic, with OpenAI and other companies constantly pushing the boundaries. He emphasized the need for speed and agility, urging employees to adopt a "scrappy" mindset and prioritize rapid development and product launches.

💬🔌 A new report suggests that the rapid growth of AI data centers in the US may be causing significant disruptions to the power grid, potentially leading to power quality issues that could damage appliances, increase fire risks, and even cause blackouts.

🎙 Recent discussions have challenged the narrative of OpenAI's sole contribution to the "scaling law" of AI, with some arguing that former OpenAI researchers observed similar phenomena while working at Baidu in 2014, suggesting that China may have been an earlier pioneer in this crucial area of AI research.

🐫 The UAE is aggressively pursuing AI leadership by liberalizing its economy, attracting global talent through relaxed immigration policies, and fostering a business-friendly environment, positioning itself as a model for economic diversification and future-proofing in the 21st century.

🤖 Nvidia is launching a new powerful computer chip, Jetson Thor, designed specifically for humanoid robots, aiming to become the dominant supplier of AI processing power to the robotics industry. Many are calling this the ‘ChatGPT‘ moment for physical AI.

📃 NotebookLM - Your personalized AI research assistant, built with Google Gemini 2.0

Pricing: On its free plan, you can have up to 100 notebooks, with each notebook containing up to 50 sources. Each of those sources can be up to 500,000 words long.

🎤 ElevenLabs - The most realistic text-to-speech speech generator and AI voiceover

Pricing: 10 minutes of ultra-high quality text to speech per month for free, 32 languages, thousands of voices

📱 Buffer - AI-based social media management tool | Grow your audience on social media by creating content to match the engagement style of each social media platform.

Pricing: Free for 10 scheduled posts per channel + 100 ideas + AI assistant

12 prompts for academic research

AI chatbots are powerful tools that can be used to accelerate many tasks in our daily lives. One of the areas in which tools like ChatGPT, Claude, or Gemini excel at, is assisting scientists with their research. Here are some very useful prompts to boost your productivity and save your time.

Literature Review Assistance: “Can you summarize the key findings of [insert academic paper title/paper’s text/article’s text]?”

Proofreading: “Can you check this paragraph for grammatical/structural errors?”

Brainstorming Ideas: “I’m working on a research project about [insert topic]. Can you suggest some potential research questions or generate some novel hypotheses?”

Methodology Suggestions: “What are some common research methodologies used in [insert field of study]? What are the state-of-the-art methods in this field?”

Writing Assistance: “Can you help me rephrase/paraphrase this sentence/paragraph in a more academic tone? Provide necessary information and context whenever needed.”

Citation Guidance: “What is the correct way to cite a web article in IEEE style?”

Concept Explanation: “Can you explain the concept of [insert complex concept] in simple terms? If possible, suggest me articles or YouTube videos, so I can better understand this concept”

Paper Structuring: “What is the typical structure of a research paper in [insert field of study] at [insert journal name]?”

Paper Restructuring: “Can you take my manuscript and rewrite it, considering the paper structure of [insert journal name]? This is how the journal needs authors to structure their papers: [insert the journal’s author guidelines/guide for authors]”

Ethical Considerations: “What are some ethical considerations when conducting research on [insert topic]?”

Theory Application: “How might [insert theory/idea] apply to a situation where [certain scenario]?”

Research Proposal Feedback: You are an elite and accomplished professor of [field of study] at an Ivy League university in the United States. You have received a graduate student’s research proposal on [proposal topic]. “Here’s a brief outline of the mentioned research proposal: [outline]”. Can you give a thorough assessment of the proposal? (Analyze its novelty, methods, objectives, writing quality, etc.). Provide useful suggestions to the student, with the potential to dramatically improve the proposal.

Empower - Staff Data Scientist - Remote

AccuWeather - Data Scientist III - Remote

Jerry - Senior Data Scientist

Your Personal AI - Machine Learning Engineer

Thank you for staying with us like always! If you are not subscribed, subscribe here for free to get more of these emails in your inbox! Cheers!