Skip to main content

Welcome to the world of GPT-3💫



The Internet is buzzing about GPT-3, OpenAI's newest AI language model.

GPT-3 is the most powerful language model ever built. This is due more than anything to its size: the model has a whopping 175 billion parameters. To put that figure into perspective, its predecessor model GPT-2—which was considered state-of-the-art and shockingly massive when it was released last year—had 1.5 billion parameters.

After originally publishing its GPT-3 research in May, OpenAI gave select members of the public access to the model last week via an API. Over the past few days, samples of text generated by GPT-3 have begun circulating widely on social media.

GPT-3’s language capabilities are breathtaking. When properly primed by a human, it can write creative fiction; it can generate functioning code; it can compose thoughtful business memos; and much more. Its possible use cases are limited only by our imaginations.

Yet there is widespread misunderstanding and hyperbole about the nature and limits of GPT-3’s abilities. It is important for the technology community to have a more clear-eyed view of what GPT-3 can and cannot do.

Comments

Popular posts from this blog

Party Pairs

  Atoms in a gas can seem like partiers at a nanoscopic rave, with particles zipping around, pairing up, and flying off again in a seemingly random fashion. And yet physicists have come up with formulas that predict this behavior, even when the atoms are extremely close together and can tug and pull on each other in complicated ways. The environment within the nucleus of a single atom seems similar, with protons and neutrons also dancing about. But because the nucleus is such a compact space, scientists have struggled to pin down the behavior of these particles, known as nucleons, in an atom’s nucleus. Models that describe the interactions of nucleons that are far apart broken down when the particles pair up and interact at close range. Now an MIT-led team has simulated the behavior of protons and neutrons in several types of atomic nuclei, using some of the most powerful supercomputers in the world. The team explored a wide range of nuclear interaction models and found, surprising...

The Next Best Thing to Jezero Crater This Side of Mars

You may not be able to travel to Jezero Crater on  , but you can visit the next best thing: Lake Salda, Turkey. Researchers are using their understanding of Lake Salda to help guide the Mars 2020 mission, which will drop the Perseverance rover into the crater to search for signs of ancient life. “One of the great things about visiting Lake Salda is it really gives you a sense of what it would have been like to stand on the shores of ancient Lake Jezero,” said Briony Horgan, a planetary scientist at Purdue University and member of the Perseverance science team. “Carbonates are important because they are really good at trapping anything that existed within that environment, such as microbes, organics, or certain textures that provide evidence of past microbial life,” said Brad Garczynski, a graduate student at Purdue who works with Horgan. “But before we go to Jezero, it is really important to gain context on how these carbonates form on Earth in order to focus our search for signs f...

Super-Resolution Imaging

Researchers at Helmholtz Zentrum München and the Technical University of Munich (TUM) have developed the world’s smallest ultrasound detector. It is based on miniaturized photonic circuits on top of a silicon chip. With a size 100 times smaller than an average human hair, the new detector can visualize features that are much smaller than previously possible, leading to what is known as super-resolution imaging. Since the development of medical ultrasound imaging in the 1950s, the core detection technology of ultrasound waves has primarily focused on using piezoelectric detectors, which convert the pressure from ultrasound waves into electric voltage. The imaging resolution achieved with ultrasound depends on the size of the piezoelectric detector employed. Reducing this size leads to higher resolution and can offer smaller, densely packed one or two-dimensional ultrasound arrays with improved ability to discriminate features in the imaged tissue or material. However, further reducing t...