Monthly Archives: September 2023
The ChatGPT Hype Is Over — Now Watch How Google Will Kill ChatGPT.
It’s happening. OpenAI’s losing the AI race.
- Remember those days when ChatGPT was everyone’s topic of conversation? Yes, you do.
- Remember those days when BeReal was everywhere? Yes, you do.
- Remember those days when Vine was the most trending app? Uh, maybe?
- What about when YikYak was everyone’s app? Yik-what?
Go back to high school. There’s always that popular girl in school for a few years. Ten years later, you’ll probably say, “Gosh, I haven’t heard that name in years.” — Read More
Large language models aren’t people. Let’s stop testing them as if they were.
When Taylor Webb played around with GPT-3 in early 2022, he was blown away by what OpenAI’s large language model appeared to be able to do. Here was a neural network trained only to predict the next word in a block of text—a jumped-up autocomplete. And yet it gave correct answers to many of the abstract problems that Webb set for it—the kind of thing you’d find in an IQ test. “I was really shocked by its ability to solve these problems,” he says. “It completely upended everything I would have predicted.”
Webb is a psychologist at the University of California, Los Angeles, who studies the different ways people and computers solve abstract problems. He was used to building neural networks that had specific reasoning capabilities bolted on. But GPT-3 seemed to have learned them for free.
… What Webb’s research highlights is only the latest in a long string of remarkable tricks pulled off by large language models.
… These kinds of results are feeding a hype machine predicting that these machines will soon come for white-collar jobs, replacing teachers, doctors, journalists, and lawyers. …But there’s a problem: there is little agreement on what those results really mean. — Read More
AI-Generated Masterpiece: 21 Savage x Travis Scott – Whiplash by @ghostwriter
Baidu CEO says more than 70 large AI language models released in China
More than 70 large artificial intelligence language models with over 1 billion parameters have been released in China, Baidu Inc (9888.HK) CEO Robin Li told an industry event in Beijing on Tuesday.
Baidu joins several other Chinese companies that launched AI chatbots last week after securing regulatory approval for mass market releases. These include facial recognition firm SenseTime (0020.HK) and AI startups Baichuan Intelligent Technology, Zhipu AI, and MiniMax. — Read More
Generative AI in Video and the Future of Storytelling (with Runway CEO Cristobal Valenzuela)
We sit down with RunwayML’s CEO Cristobal Valenzuela to discuss the incredible tools they’re bringing to film and video creators (including last year’s Best Picture “Everything Everywhere All at Once” from A24), and the history + current state of the “visual” branch of generative AI. We cover how they’ve gone to market with both creators and enterprises, the potential for much more radical future use cases, and the company’s recent $141m strategic raise from Google, Nvidia + Salesforce and the context of the current AI fundraising landscape. Tune in! — Read More
Andrew Ng: Opportunities in AI – 2023
Why “AI” can’t succeed without APIs
Mega tech trends like the cloud, the mobile phone era, metaverse and now AI all depend on enabling technologies sitting right beneath the surface hidden from nearly everyone’s view. Their structural integrity depends on the flawless operation of those enabling technologies, which in many cases are Application Programming Interfaces (APIs). As such, their success depends on API adoption. Nowhere is this truer than in the rapid proliferation of AI technologies, like generative AI, which require a simple and very easy-to-use interface that gives everyone access to the technology. The secret here is that these AI tools are just thin UIs on top of APIs that connect into the highly complex and intensive work of a large language model (LLM).
It’s important to remember that AI models don’t think for themselves, they only appear to be so that we can interact with them in a familiar way. APIs are essentially acting as translators for AI platforms as they’re relatively straightforward, highly structured and standardized on a technological level. What most people think of as “AI” should be viewed through the lens of an API product; and with that mindset, organizations can best prepare for what potential use cases are possible and how to ensure their workforces have the skills to put them into action. — Read More
New York police will use drones to monitor backyard parties this weekend, spurring privacy concerns
Those attending outdoor parties or barbecues in New York City this weekend may notice an uninvited guest looming over their festivities: a police surveillance drone.
The New York City police department plans to pilot the unmanned aircrafts in response to complaints about large gatherings, including private events, over Labor Day weekend, officials announced Thursday. — Read More
How to Fine-Tune Llama2 for Python Coding on Consumer Hardware
Our previous article covered Llama 2 in detail, presenting the family of Large Language models (LLMs) that Meta introduced recently and made available for the community for research and commercial use. There are variants already designed for specific tasks; for example, Llama2-Chat for chat applications. Still, we might want to get an LLM even more tailored for our application.
Following this line of thought, the technique we are referring to is transfer learning. This approach involves leveraging the vast knowledge already in models like Llama2 and transferring that understanding to a new domain. Fine-tuning is a subset or specific form of transfer learning. In fine-tuning, the weights of the entire model, including the pre-trained layers, are typically allowed to adjust to the new data. It means that the knowledge gained during pre-training is refined based on the specifics of the new task.
In this article, we outline a systematic approach to enhance Llama2’s proficiency in Python coding tasks by fine-tuning it on a custom dataset. — Read More
Rick's Cafe AI 