The Year of AI Agents

The Year of AI Agents

Imagine personal bots acting on your behalf – from searching for vacations and booking flights and hotels, to conducting and summarizing sales calls, selling products or services, and maintaining customer relationships. Tech giants like Microsoft and Google are developing these AI agents, capable of automating entire processes from start to finish.
What does this mean for your job?
The world is about to change dramatically in the next two years.
I’m here to help organizations train their employees for AI technology. Connect with me or share your manager’s details, and I’ll be happy to get in touch.

 

 Welcome to the Duplicated World

Polish startup, Clone Robotics, is taking robotics to the next level. They’ve developed Clone Alpha, a robot with skeletal, muscular, and nervous systems similar to ours. It walks like us, including hand and finger movements, can fold laundry, assemble furniture, and learn new skills. The first 279 robots will be on sale next year – and I want one for chores!

Decoding our Genome

In December 2024, InstaDeep unveiled Nucleotide Transformers (NT), a new model designed to decode DNA sequences using Transformer architecture. Trained on a vast dataset of over 3,200 human genomes, NT models excel at tasks like splice site prediction and enhancer activity modeling. They outperform competitors and can predict the impact of genetic mutations. Available on platforms like Hugging Face and through InstaDeep’s DeepChain, NT is a game-changer for personalized medicine and genomic research, supported by global collaborations. Learn more.

Forever Young

Scientists are exploring the secrets of youth with AgeXtend, a new AI-driven platform that analyzes bioactivity data to discover anti-aging molecules known as geroprotectors. Soon, we may be able to stop aging and remain young forever.

12 Announcements from OpenAI

OpenAI unveiled 12 new features for ChatGPT, including:

  • o1-Preview Model: Limited to 40 prompts per week for paid subscribers. Unlimited access is available for $200/month with the new Pro plan. o1 also boasts multi-modal capabilities, accepting visual input.

  • Customizable Training: Train models on specific data in fields like medicine, economics, nutrition, etc.

  • Sora – Text-to-Video Generator: Finally released! While currently unavailable for new sign-ups due to high demand, Plus subscribers can create up to 50 five-second videos per month at 720p with watermarks. The new $200/month Pro subscription allows for up to 500 videos (up to 20 seconds each) at 1080p without watermarks, with high-speed generation initially and then unlimited lower-speed generation.

  • Canvas for Free Users: The Canvas feature is now open to free users. (I’ve created a video tutorial on how to use it.)

  • Siri Integrates ChatGPT: iPhone 15 and 16 users can now access ChatGPT’s full functionality, including text and image generation and multilingual conversations, directly through Siri.

  • Live Video Chat: OpenAI released real-time video chat, available for paid subscribers.

  • Project Folders: Organize your projects with dedicated folders for uploading source materials and linking related conversations.

  • Web Search for Free Users: OpenAI opened web search capabilities to free users.

  • ChatGPT on WhatsApp: Add +1(800)242-8478 to your contacts and chat with ChatGPT via WhatsApp.

  • New o3 Model: A significantly more advanced model than o1.

  • o1 Upgrade: Improved performance on complex tasks and API access for developers to connect with other tools.

  • ChatGPT for Mac Apps: Mac users can now access ChatGPT directly from various Mac applications.

Alibaba vs. OpenAI

Alibaba releases an ‘open’ challenger to OpenAI’s o1 reasoning model.
Alibaba has introduced the QwQ-32B-Preview, a 32.5 billion-parameter AI model designed to enhance logical reasoning and problem-solving capabilities. This model can process prompts up to 32,000 words and has demonstrated superior performance on benchmarks like AIME and MATH, surpassing OpenAI’s o1-preview and o1-mini models in specific areas.

Google’s New Video Model for Businesses

Google announced Veo, a cutting-edge video generation model. Veo empowers users to create personalized, high-quality videos, including characters, unique visual effects, advertisements, animated films, music videos, and more. Its intuitive interface provides users with control over the creative process, customization options, and tailoring content to their specific needs.
Currently in testing with select users, Google aims to gather feedback, refine the model, and make it widely available in the future.

60 of Google’s biggest AI announcements in 2024

Google AI had a busy year in 2024, releasing new features and updates across its products. From the launch of Gemini 1.5 and the integration of AI into Google Search to the development of new AI tools for health and education, Google AI made significant strides in 2024. What next? Google AI is poised to continue its momentum, with new advancements in areas like quantum computing and agentic AI.

 

100,000 New Jobs

SoftBank CEO Masayoshi Son is poised to announce a $100 billion investment in the US over the next four years, focusing on creating 100,000 new jobs in AI. While 100,000 sounds impressive, forecasters predict the disappearance of nearly 2 billion jobs by 2030 – a staggering number!

Preventative Medicine a Step Closer

Pillbot, a 13mm pill, travels through the stomach to observe internal bodily processes. This technology could revolutionize how we gather health information, enabling early disease detection and treatment before symptoms appear. In this video, Nick swallows a Pillbot to demonstrate its functionality and discovers something unexpected.

AI in Mammography

A new study by DeepHealth shows that women undergoing AI-assisted mammograms have a 21% higher chance of early breast cancer detection compared to traditional methods.

Harvard Releases AI Training Dataset

Harvard University released a dataset for training AI models containing over a million public domain books. Five times larger than the renowned Books3 dataset used to train models like Meta’s Llama, this project is funded by OpenAI and Google. It will include books scanned by Google Books and not protected by copyright, providing a rich variety of texts in various languages, including classics by Shakespeare, Dickens, and Dante, as well as textbooks.

AI-Powered Oreo Cookies

Oreo is using AI to create new and exciting cookie variations, believing it can enhance taste, texture, and appearance, offer a wider variety, and create unique experiences by analyzing data, identifying patterns, and developing innovative recipes. Staying relevant means catering to evolving consumer needs and adapting to market trends.

AI Quick Updates

  • GenChess: Google created a new online chess game where players design custom chess pieces using AI.

  • Free Harvard Courses: A list of 54 free top-tier Harvard courses.

  • Smart Swim 2 Goggles: Smart swimming goggles displaying heart rate, swim time, pace, and pulse in real-time.

  • AgeXtend Platform: Scientists are searching for the fountain of youth with AgeXtend, an AI-powered platform utilizing biological activity data to identify anti-aging molecules. I believe we’ll soon be able to stop aging and stay young forever. (Link to article.)
  • Gemini 2.0: Google released Gemini 2.0, enabling real-time voice and video conversations and leveraging Claude for enhanced performance.

  • AI-Powered DUI Detection: Southwest England will host a pilot program testing AI’s ability to identify drivers under the influence of drugs or alcohol through cameras.

  • Claude Update: Upload up to 20 files (max 30MB each) to Claude.

 

AI productive tools:

OpenScholar – An AI platform that synthesizes open access computer science studies to help you find, summarize, and learn about 1M+ research papers.

Doctranslate.io – Seamlessly translates documents of any type into dozens of languages faster, cheaper and better

Spotterstudio – This tool helps YouTube creators create more winning videos through an integrated suite of ideation tools.

ComicsMaker is an AI-powered comic generator

Desmos – is a learning and teaching platform with hundreds of free, interactive K–12 lessons

לפרטים נוספים

נא מלאו את הטופס ואצור איתכם קשר בהקדם 

Skip to content