Since ChatGPT launched in November, small businesses have battled a broader recession in the tech industry. No week goes by without someone launching a “generative” artificial intelligence (AI) based on a “basic” model. This model is a vast and complex algorithm that gives ChatGPT and other AI intelligence. On February 24th, Facebook parent company Meta released a model called LLaMA. Tesla and his Twitter president, Elon Musk, reportedly want to create an AI that is “less awake” than ChatGPT. Curated by British entrepreneur Ben Tossel, one of his catalogs now includes Isaac Editor, among others. Helping students write essays), Pickaxe (analyzing their own documents), Ask Seneca (answering questions based on the writings of Stoic philosophers), ChatGPT can get a lot of buzz, 100 million There are more users than people, and it can be a hot topic. Still, Tossell’s database his tip is that the real action in generative AI isn’t all sorts of very chatty services enabled by underlying models.
Since ChatGPT launched in November, small businesses have battled a broader recession in the tech industry. No week goes by without someone launching a “generative” artificial intelligence (AI) based on a “basic” model. This model is a vast and complex algorithm that gives ChatGPT and other AI intelligence. On February 24th, Facebook parent company Meta released a model called LLaMA. Tesla and his Twitter president, Elon Musk, reportedly want to create an AI that is “less awake” than ChatGPT. Curated by British entrepreneur Ben Tossel, one of his catalogs now includes Isaac Editor, among others. Helping students write essays), Pickaxe (analyzing their own documents), Ask Seneca (answering questions based on the writings of Stoic philosophers), ChatGPT can get a lot of buzz, 100 million There are more users than people, and it can be a hot topic. Still, Tossell’s database his tip is that the real action in generative AI isn’t all sorts of very chatty services enabled by underlying models.
Each model is trained on a set of text, images, sound files, or other data. This allows them to interpret instructions in natural language and respond with text, art, or music. Such systems have existed for some time, but they needed a consumer-facing service like ChatGPT to capture the world and the imagination of investors. As his Mike Volpi of venture capital (VC) firm Index Ventures puts it, this is exactly what his tech backers, hurt by the cryptocurrency crash and an empty metaverse, were looking for for the next big thing. It happened then. Moreover, more than web browsers and smartphones, the underlying model makes it easy to build new services and applications on top of it. “You can open up your laptop, get an account, and start interacting with models,” says Steve Loughlin of another of his VC firms, Accel.
Hi!reading premium articles
Each model is trained on a set of text, images, sound files, or other data. This allows them to interpret instructions in natural language and respond with text, art, or music. Such systems have existed for some time, but they needed a consumer-facing service like ChatGPT to capture the world and the imagination of investors. As his Mike Volpi of venture capital (VC) firm Index Ventures puts it, this is exactly what his tech backers, hurt by the cryptocurrency crash and an empty metaverse, were looking for for the next big thing. It happened then. Moreover, more than web browsers and smartphones, the underlying model makes it easy to build new services and applications on top of it. “You can open up your laptop, get an account, and start interacting with models,” says Steve Loughlin of another of his VC firms, Accel.
Money is pouring into business. In January, it was reported that Microsoft poured $10 billion into OpenAI, the startup behind ChatGPT, on top of an earlier $1 billion investment. Pete Flint, of another VC firm, NfX, now counts over 500 generative AI startups. They’ve raised over $11 billion in total to date, which doesn’t include his OpenAI (see chart). Volpi talks about the “Cambrian Explosion”.
Click image to enlarge
So which generative AI platforms will make the big bucks? So far, this has been a hot topic in the tech world. “It is not clear whether long-term winner-takes-all dynamics exist in generative AI,” Martin Casado and colleagues at Andreessen Horowitz, his VC firm, said in a recent blog post. said in Many startups have made offers to me. While OpenAI’s proprietary models such as his GPT-3.5 are ahead, open-source alternatives aren’t far behind.
Generative AI is also creeping into the legal minefield. Models often get things wrong. And they can go off the rails. Sydney, Microsoft’s OpenAI-powered chatbot for the Bing search engine, has insulted several users and confessed love to at least one (and has since been cracked down). AI platforms may not enjoy the legal protections that protect social media. Without asking permission or paying compensation, copyright holders of web-based content on which existing models are trained are armed. Photo repository Getty Images and individual artists have filed lawsuits against AI art generators such as Stable Diffusion. Stable Diffusion says: We are reviewing the documents and will respond accordingly. News outlets are also concerned about AI reading text (see next article).
OpenAI has already downplayed the launch of GPT-4, a long-awaited update to its underlying model, later this year. VC-type appetite for generative AI is not diminishing. For the more risk-averse investor, the safest bet right now is to rely on a provider that provides enough processing power to train and run the underlying model. Nvidia, which designs chips that help AI applications, is up 60% this year. Landlords of cloud computing services and data centers are doing the same. No matter which AI platform comes out on top, you can’t go wrong selling pickaxes and shovels in the Gold Rush.
©️ 2023, The Economist Newspaper. all rights reserved.
Published under license by The Economist. Original content is available at www.economist.com.
