OpenAI today launched the OpenAI Startup Fund, a $100 million fund to -- in the words of OpenAI -- "help AI companies have a profound, positive impact on the world." The fund is managed by OpenAI, with investment from Microsoft and other partners, and OpenAI says that companies selected for it will get early access to future OpenAI systems, support from OpenAI's team, and credits on Microsoft Azure. According to Sam Altman, CEO of OpenAI and the former president of Y Combinator, the OpenAI Startup Fund will make "big, early bets" on a relatively small number of companies, likely no more than 10. It'll look to partner with early-stage startups in fields where AI can have a "transformative" effect -- like health care, climate change, and education -- and where AI tools can empower people by helping them be more productive, like personal assistance and semantic search. "We think that helping people be more productive with new tools is a big deal. And we can imagine brand new interferences that weren't possible a year ago," Altman said.

Microsoft has officially introduced its first GPT-3-powered feature in a customer product, eight months after it exclusively licensed the sophisticated OpenAI language model. The tech giant has announced at the virtual Build developers conference that it's integrating GPT-3 in Power Apps, which even people with no coding experience can use to build business productivity apps. With the new features in place, Power Apps will be even easier to use -- in fact, it'll give users the power to code by using plain conversational language. GPT-3 is the largest language model ever trained and is capable of generating text so human-like, it could write believable fake news. Microsoft invested $1 billion in OpenAI back in 2019 and got access to the language tech for its own use and for its Azure cloud customers.

Build Any souls wondering what Microsoft would do with its GPT-3 investment have been given an answer with a Power Fx update lightly seasoned with the AI tech. Microsoft gained exclusive rights to use OpenAI's GPT-3 in September last year, allowing it to embed the text-and-code-generating machine-learning model into its own products. Available in preview from next month, the technology was shown off at Microsoft's Build 2021 shindig today, and represents the latest attempt by the Windows giant to get folks from low code to no code and bring its Power platform closer to the masses. Looking initially like a jumped-up version of IntelliSense, the technology attempts to parse natural language entered by the user and generate the corresponding Excel-like language of Power Fx to perform the requested task. The idea is that you type in something like, "show me the readers who commented at the weekend," and it should generate the formulas to retrieve that information.

In September 2020, Microsoft purchased an exclusive license to the underlying technology behind GPT-3, an AI language tool built by OpenAI. Now, the Redmond, Washington-based tech giant has announced its first commercial use case for the program: an assistive feature in the company's PowerApps software that turns natural language into readymade code. The feature is limited in its scope and can only produce formulas in Microsoft Power Fx, a simple programming language derived from Microsoft Excel formulas that's used mainly for database queries. But it shows the huge potential for machine learning to help novice programmers by functioning as an autocomplete tool for code. There's a million-developer shortfall in the US alone," Charles Lamanna, CVP of Microsoft's Low Code Application Platform, tells The Verge. "So instead of making the world learn how to code, why don't we make development environments speak the language of a normal human?" Microsoft has been pursuing this vision for a while through Power Platform, its suite of "low code, no code" software aimed at enterprise customers. These programs run as web apps and help companies that can't hire experienced programmers tackle basic digital tasks like analytics, data visualization, and workflow automation. GPT-3's talents have found a home in PowerApps, a program in the suite used to create simple web and mobile apps. Lamanna demonstrates the software by opening up an example app built by Coca-Cola to keep track of its supplies of cola concentrate. Elements in the app like buttons can be dragged and dropped around the app as if the users were arranging a PowerPoint presentation. But creating the menus that let users run specific database queries (like, say, searching for all supplies that were delivered to a specific location at a specific time) requires basic coding in the form of Microsoft Power Fx formulas. "This is when it goes from no code to low code," says Lamanna. "You go from drag and drop, click click click, to writing formulas.

This is a basic Python shell (really, it's a fancy wrapper over the system shell) that takes a task and asks OpenAI for what Linux bash command to run based on your description. For safety reasons, you can look at the command and cancel before actually running it. To be clear, I'm not trying to convince you that having an AI model figure out what Linux command to run based on your written description is a good idea, but the commands that it generates are, well - watch the video if you want to see. There are several pre-canned ways of interacting with the models that OpenAI provides (the "GPT" models): completing a provided fragment, answering a question, generating "ideas" from a topic, summarizing a passage, etc. This shell uses the question-and-answer format and provides the model with an "example context" and examples of input and output.

When OpenAI demonstrated a powerful artificial intelligence algorithm capable of generating coherent text last June, its creators warned that the tool could potentially be wielded as a weapon of online misinformation. Now a team of disinformation experts has demonstrated how effectively that algorithm, called GPT-3, could be used to mislead and misinform. The results suggest that although AI may not be a match for the best Russian meme-making operative, it could amplify some forms of deception that would be especially difficult to spot. Over six months, a group at Georgetown University's Center for Security and Emerging Technology used GPT-3 to generate misinformation, including stories around a false narrative, news articles altered to push a bogus perspective, and tweets riffing on particular points of disinformation. "I don't think it's a coincidence that climate change is the new global warming," read a sample tweet composed by GPT-3 that aimed to stoke skepticism about climate change.

A new report lays out the ways that cutting-edge text-generating AI models could be used to aid disinformation campaigns. Why it matters: In the wrong hands text-generating systems could be used to scale up state-sponsored disinformation efforts -- and humans would struggle to know when they're being lied to. How it works: Text-generating models like OpenAI's leading GPT-3 are trained on vast volumes of internet data, and learn to write eerily life-like text off human prompts. What they found: While "no currently existing autonomous system could replace the entirety of the IRA," algorithmically based tech paired with experienced human operators produces results that are nothing less than frightening. What to watch: While OpenAI has tightly restricted access to GPT-3, Buchanan notes that it's "likely that open source versions of GPT-3 will eventually emerge, greatly complicating any efforts to lock the technology down."

Recent research has shown that numerous human-interpretable directions exist in the latent space of GANs. In this paper, we develop an automatic procedure for finding directions that lead to foreground-background image separation, and we use these directions to train an image segmentation model without human supervision. Our method is generator-agnostic, producing strong segmentation results with a wide range of different GAN architectures. Furthermore, by leveraging GANs pretrained on large datasets such as ImageNet, we are able to segment images from a range of domains without further training or finetuning. Evaluating our method on image segmentation benchmarks, we compare favorably to prior work while using neither human supervision nor access to the training data. Broadly, our results demonstrate that automatically extracting foreground-background structure from pretrained deep generative models can serve as a remarkably effective substitute for human supervision.

That element of randomness is partially why GPT-3--or its less powerful predecessor, GPT-2--is taking time to gain widespread commercial traction as a tool to power chatbots or auto-generate ads. After nearly a year of experimentation, however, how such a technology might be tamed for marketing purposes is beginning to take shape. Working through OpenAI's closely guarded API program, startups and agency technologists have reined in GPT-3's more eccentric tendencies, which can range from nonsensical prose to inappropriate or explicit content, in order to put it to use for rote performance marketing tasks like A/B testing endless variations of a digital ad, generating product descriptions or assigning email subject lines. Meanwhile, other companies are capitalizing on GPT-3's stranger side for creative tools. While still nascent, projects like these offer a glimpse into a future where humans might work hand in hand with generative AI on creative copywriting and the give-and-take forces that might define such a relationship.

We investigate statistical properties of a likelihood approach to nonparametric estimation of a singular distribution using deep generative models. More specifically, a deep generative model is used to model high-dimensional data that are assumed to concentrate around some low-dimensional structure. Estimating the distribution supported on this low-dimensional structure such as a low-dimensional manifold is challenging due to its singularity with respect to the Lebesgue measure in the ambient space. In the considered model, a usual likelihood approach can fail to estimate the target distribution consistently due to the singularity. We prove that a novel and effective solution exists by perturbing the data with an instance noise which leads to consistent estimation of the underlying distribution with desirable convergence rates. We also characterize the class of distributions that can be efficiently estimated via deep generative models. This class is sufficiently general to contain various structured distributions such as product distributions, classically smooth distributions and distributions supported on a low-dimensional manifold. Our analysis provides some insights on how deep generative models can avoid the curse of dimensionality for nonparametric distribution estimation. We conduct thorough simulation study and real data analysis to empirically demonstrate that the proposed data perturbation technique improves the estimation performance significantly.