- Synthetic
- Posts
- Humanoid Robot Painting May Sell For $120K or More
Humanoid Robot Painting May Sell For $120K or More
Plus, What Reasoning AIs Mean For Business And Beyond
Subscribe to Synthetic
The week’s most interesting and relevant AI news and analysis
This Week in AI
Federal crime busters’ big bet on AI is paying off. The US Treasury used machine learning to sift through oceans of data and recover $1 billion of check fraud in fiscal 2024, tripling the amount they recovered the previous year. They credit AI with recovering a total of $4 billion of overall fraud this year, a six-fold increase from 2023. The Treasury Department delivers about 1.4 billion payments each year valued at $7 trillion to 100 million people. Online payment fraud is predicted to exceed $362 billion by 2028 (Source: Juniper Research). 💰
Ai-Da, an experimental AI-powered robot is the first of its kind to have its artwork sold by a major auction house. Sotheby’s is accepting bids starting at $120,000 for the abstract portrait of Alan Turing, titled “AI God.” The robot takes visual input, generates images, and then uses its arms to create the image on paper with paint. 🎨
The Microsoft Threat Analysis Center reported that Russian operatives have created AI-enhanced deepfake videos designed to portray Vice President Harris in an unfavorable light. Less than two weeks before the election, fake videos have begun circulating that show Harris making a “crass reference” to assassination attempts against Donald Trump and accusing her of illegally poaching in Zambia. Other disinformation videos target Governor Tim Walz. The National Intelligence Council expects such efforts to intensify in the coming days. Chinese and Iranian influence operations have also been detected. The key lesson: Think critically about what you see online. It could well be manipulation by hostile foreign entities. In related news, Warren Buffett says he doesn’t endorse any candidate or investment product despite numerous deepfakes that claim otherwise. Don’t forget to vote. 🗳️
I’m worried about people impersonating me and that’s why we put that on the Berkshire Web site. Nobody should believe anybody saying I’m telling them how to invest or how to vote.”
Video: Attention in Transformers, Visually Explained
3Blue1Brown is a fabulous resource for learning about science and AI. This episode, published earlier this year, explains how attention mechanisms work. These lie at the heart of all large language models and are essential to their operation. You’ll learn about embeddings, attention, and the magic of how modern AI works. (26 mins).
If you’re new to AI, you may want to watch this episode first on how large language models work. (27 mins)
AI Tech and Innovation
So-called ‘Agentic AI’ has been the talk of the AI research community for some time. This week, Anthropic became the first frontier lab to release an AI agent. Anthropic’s new Computer Use capability enables its Claude 3.5 Sonnet model to interpret what’s on a computer screen, select buttons, enter text, navigate websites, and perform complex tasks requiring software interactions or internet browsing. Anthropic demonstrated Claude’s new skills in this 2-minute video.
Note from Synthetic: OpenAI and Google DeepMind are believed to be working on agentic AI capabilities. Anthropic’s announcement may accelerate plans at those two leading labs to release similar capabilities in the coming weeks.
Speaking at the TED AI Conference this week in San Francisco, OpenAI leading research scientist Noam Brown discussed how the System Two thinking showcased in their new o1 model will yield significant benefits in the future. Since the invention of the transformer architecture in 2017, most advances in AI have come from scaling up models and stuffing them with more and more intelligence at training time. OpenAI has said that they don’t anticipate much more capability will come from this type of System One scaling, and they see deep reasoning as the primary way to scale capabilities in the future. Deep reasoning occurs at test time, during inference. The o1 model takes time to reason through complex problems, improving accuracy and expanding its problem-solving capabilities. This extra time comes at a cost. The o1 preview model costs $15 per million input tokens and $60 per million output tokens, significantly more than other models, reflecting the increased computing power required to run such models. ⚡️
Plagiarism, copyright violations, and misinformation are major downsides of the recent AI revolution. We need an easy way to tell when text, audio, images, and video have been generated by AI. This week, Google DeepMind open-sourced its SynthID tool for identifying AI-generated text, music, and other data types. The new tool is highly accurate, though it's not foolproof, according to VP of Research Pushmeet Kohli. Details of their work were published in Nature and Google DeepMind published a blog to explain how SynthID works. Last year, Google, OpenAI, Meta, and other AI companies voluntarily committed to developing tools for detecting AI-generated text.
AI Insights
Reasoning AIs will be transformational
OpenAI’s new o1 model has introduced a new level of capabilities to AI—the ability to break down complex problems and ‘think’ through them step-by-step, refine their thinking, try different strategies, and recognize and correct mistakes to reason a solution and provide a final answer. These reasoning skills are vital in fields including law, medicine, finance, strategic planning, scientific research, business decision-making, customer support, and software development. Businesses should lean into such technology but remain cautious. Over reliance on AI vs. human reasoning may make your company less resilient, not more.
While 96% of top executives say they expect AI tools to increase productivity (Source: Upwork survey of 2,500 knowledge workers in the US, UK, Australia, and Canada), 77% of employees say AI tools decreased their productivity and added to their workload. As employees struggle to keep up with the bewildering array of new AI tools coming to market and how to get the best out of new features, 61% of people believe AI at work increases their chance of burnout, with 87% of people under 25 reporting the same (Source: Resume Now survey of 1,150 Americans).
Hollywood creative types, musicians, and authors have awakened to the fact that a wave of AI-powered change is coming their way. Their open letter calls attention to what they call the ‘major’ threat of allowing generative AI to train on their work. Over 19,000 people have signed the statement: “The unlicensed use of creative works for training generative AI is an unjust threat to the livelihoods of the people behind those works, and must not be permitted.” The bloated U.S. Movie and TV industry employs tens of thousands of people—carpenters, set designers, make-up artists, actors, directors, and many more. It faces ever-increasing cost pressures as viewers switch to streaming platforms and international competition grows.
Synthetic’s Take: Expect many more similar statements, strikes, and protests as AI’s capabilities grow and job losses increase across various industries.
Toolkit for the Future
Learn AI in 5 Minutes a Day
AI Tool Report is one of the fastest-growing and most respected newsletters in the world, with over 550,000 readers from companies like OpenAI, Nvidia, Meta, Microsoft, and more.
Our research team spends hundreds of hours a week summarizing the latest news, and finding you the best opportunities to save time and earn more using AI.
Maintain healthy calendar habits, improve your productivity, optimize cross-team meetings, boost collaboration, and improve work-life balance. Reclaim schedules 1:1s and defends your calendar so you can focus. Syncs multiple calendars and integrates with Google Calendar, Zoom, Slack, and HubSpot.
Take customer support to the next level with powerful AI features including chatbots, workflow automation, emotion insights, knowledge bases, data-drive insights, real-time reporting, and empathy coaching to help your support teams increase productivity and customer satisfaction.
Incorporate your company, access one-click growth tools, stay compliant, and manage everything your business needs — all online, from anywhere. Launch your U.S. business in minutes with no paperwork or legal headaches.
Need to expand internationally, but not sure where to start? Oyster makes it easy to find, hire, and retain local talent. They handle the details (country-specific labor laws, international tax laws, compliance, and global payroll) so you can focus on finding the right talent and ramping their impact.
Build intelligent personal shopping experiences by adding the Manifest AI chatbot to your site. Help shoppers find what they need, faster. Double add-to-cart and conversion rates and get 25% higher AOV. Easy Shopify and help desk integration. Free 14-day trials.
Hire top talent with industry-specific expertise to build world-class engineering teams and solve complex business problems. AI-assisted sourcing, vetting, and hiring delivers the right talent, at the right time, globally!