Generative AI appeared at the Davos Forum. OpenAI CEO said that AI will move towards the next era.

In the past year, generative AI has emerged strongly.

From text to images to videos and even codes, almost any generative AI you can think of can help you do it.

No, the 2023 World Economic Annual Summit, the Davos Forum, also named and praised the brilliant achievements of generative AI in 2022.

This article refers to generative AI as "a game-breaker that society and enterprises need to seriously deal with", which is enough to show its importance.

In addition, the forum also invited Sam Altman, CEO of OpenAI, to attend and deliver the closing keynote speech of "AI Moving towards the Next Era". A series of topics including how the future development of AI will help the economy and society were discussed.

##This time the conversation with Altman is Reid Hoffman, a well-known American company and co-founder of LinkedIn.

We have intercepted a short snippet of the conversation and interview, and the link to the full interview is also posted below.

Reid Hoffman: So I think one of the things that a lot of people are interested in is that very large models will be created based on the API, so What are the real business opportunities? What are the ways to look forward? Given that the API will be available to multiple players, how do you create a unique business?

Sam Altman: Yes. So I think so far we've entered a realm of endless possibilities where you can do a lot of things with models that were complicated in the past. But I suspect that with the quality of language models we see in the next few years, the search product will pose a serious challenge to Google for the first time. Our ChatGPT was ridiculed before, but now it demonstrates capabilities that no company can ignore.

Altman believes that artificial intelligence is the basic platform for the development of all science and technology, from the huge leap from large language models to multi-modal applications that switch between images and language, to the significant expansion of Applications of scientists' abilities from which many advances will be made across all industries.

## Video link: https://youtu.be/WHoWGNQRXb0

Full interview link: https://greylock.com/greymatter/sam-altman-ai-for-the-next-era/

###The last article is still here Summarizes the development history of generative AI over the years. ############Let’s take a look at the previous set of pictures. From 2014 to 2022, AI has made a leap in image quality. These pictures are all people generated by AI models that do not exist in reality. ##################The super power of generative AI#########The Generative Pre-trained Transformer (GPT) is a large language model ( LLM), which uses deep learning to generate human-like text. ######

Despite the current market downturn and layoffs in the tech industry, generative AI companies continue to gain attention from investors.

For example, Stability AI and Jasper recently raised US$101 million and US$125 million respectively. Investors such as Sequoia Capital believe that the field of generative AI can generate trillions of dollars in economic value.

With the release of new models such as Stable Diffusion and ChatGPT, generative AI has become a key topic for technology experts, investors, policymakers and society at large. hot topics.

Generative AI is not a new concept. The machine learning technology behind generative AI has been developing continuously over the past decade.

Generative AI training models work by learning from large data sets and using this knowledge to generate new data that is similar to the examples in the training data set.

This is typically done using a machine learning algorithm called a generative model. There are many different types of generative models, each using a different method to generate new data.

Some common generative model types include generative adversarial networks (GAN), VAEs, and autoregressive models.

For example, a generative model trained on a dataset of face images might learn the general structure and appearance of faces and then use this knowledge to generate new, previously unseen but Faces that look real and believable.

Generative models are used in a variety of applications, including image generation, natural language processing, and music generation. They are particularly useful for tasks where manually generating new data is difficult, such as creating new designs for products or generating realistic-sounding speech.

OpenAI’s latest version, ChatGPT, has caused a stir and attracted 1 million users in just five days, and has been described as a breakthrough across a wider range of tasks.

Use cases currently being discussed include new architectures for search engines, explaining complex algorithms, creating personalized treatment bots, helping build apps from scratch, explaining scientific concepts, and more.

Text-to-image programs like Midjourney, DALL-E, and Stable Diffusion have the potential to change the way art, animation, games, movies, architecture, and more are rendered.

Based on a new era of human-machine collaboration, optimists claim that generative AI will aid the creative process of artists and designers as generative AI systems will enhance reality. Having tasks accelerates ideation and radically speeds up the creative phase.

In addition to the creative space, generative AI models have transformative capabilities in complex scientific fields such as computer engineering.

For example, Microsoft-owned GitHub Copilot is based on OpenAI’s Codex model and can suggest code and assist developers in automating their programming tasks.

The system is cited as automating up to 40% of developer code, significantly increasing workflow.

Potential and Unknown Risks

While generative AI has people excited about the creativity it brings, there are also concerns about the impact of these models on society. Influence.

Digital artist Greg Rutkowski is worried that the Internet will be flooded with artwork that is indistinguishable from his own work, simply by telling the system to copy the artwork in his unique style.

Art professor Carson Grubaugh shares the same concerns and predicts that much of the creative workforce, including commercial artists working in the entertainment, video games, advertising and publishing industries, could be put out of work by generative AI models.

In addition to their profound impact on tasks and jobs, generative AI models and related externalities have caused alarm in the AI ​​governance community.

One of the problems with large language models is their ability to generate false and misleading content.

Meta’s Galactica — a model trained on 48 million scientific articles that claims it can summarize academic papers, solve mathematical problems, and write scientific code — went live less than three days ago It was withdrawn because the scientific community found that it had misunderstood students and produced incorrect data and knowledge.

In addition to the ability of bots that pass the Turing test to exhibit intelligent behavior similar to or indistinguishable from humans, such capabilities may be abused to generate disinformation across platforms and ecosystems.

Large models continue to be trained on massive data sets represented in books, articles, and websites, which may be biased in ways that are difficult to fully filter.

Although harmful and inauthentic outputs have been greatly reduced through the use of reinforcement learning with human feedback (RLHF) in the case of ChatGPT, OpenAI admits that their models can still be malicious and biased Output.

The above is the detailed content of Generative AI appeared at the Davos Forum. OpenAI CEO said that AI will move towards the next era.. For more information, please follow other related articles on the PHP Chinese website!