Artificial Intelligence: This artificial intelligence from Deep Science “simulates” savings and predicts which startups will receive funding!





Research in the area of ​​machine learning and artificial intelligence, which is now a key technology in virtually every industry and business, is too voluminous for anyone to read it completely. This section aims to bring together some of the most relevant recent discoveries and work – especially in the field of artificial intelligence, but not only – and explain why they are important.

This week, in the field of AI, scientists have conducted a fascinating experiment predict how “market-driven” platforms, such as food delivery and shared travel companies, affect the overall economy when optimized for different purposes, such as maximizing revenue. Elsewhere, demonstrating the versatility of AI, a team at ETH Zurich has developed a system capable of reading the height of trees from satellite imagery, while another group of researchers has tested a system to predict the success of a startup from public web data.

Work on the market-driven platform is driven by Salesforce’s AI Economist, an open source search environment for understanding how AI could improve economic policy. In fact, some of the researchers behind AI Economist participated in this new work, which was detailed in a study published last March.

As the co-authors told TechCrunch by email, the goal was to study two-sided markets such as Amazon, DoorDash, Uber, and TaskRabbit, which enjoy greater market power due to increased demand and ‘offer. Using reinforcement learning, a type of AI system that learns to solve a multilevel problem through trial and error, researchers trained a system to understand the impact of interactions between platforms (e.g., Lyft) and consumers. (e.g. users).

“We use reinforcement learning to reason about how a platform would work given the different design goals … [Notre] The simulator allows you to evaluate reinforcement learning policies in a variety of contexts under different model goals and assumptions, “the co-authors told TechCrunch by email.” We explored a total of 15 different market parameters, that is, one combination of market structure, knowledge of sellers of buyers, the intensity of the shock. [économique] and design purpose. »

Using their AI system, the researchers concluded that a platform designed to maximize revenue tends to raise rates and extract more profits from buyers and sellers during economic shocks, to the detriment of social welfare. When setting platform rates (e.g., due to regulation), they found that the incentive to maximize a platform’s revenue generally aligns with the welfare considerations of the global economy.

The results may not be puzzling, but the co-authors believe that the system, which they plan to open, it could serve as a basis for a company or a politician to analyze an economy platform under different conditions, designs and regulatory considerations.

“We adopt reinforcement learning as a methodology to describe the strategic operations of platform companies that optimize their prices and their concordance in response to changes in the environment, whether it is the economic shock or a certain regulation,” they said. add. “This may provide new insights into the economies of platforms that go beyond this work or those that can be generated analytically.”

Researchers at Skopai, a startup that uses artificial intelligence to characterize companies based on criteria such as technology, market and finance, say they can predict a startup’s ability to attract investment through public data. Based on data from startup websites, social media, and business logs, the co-authors claim that they can achieve prediction results “comparable to those that also use structured data available in private databases.” »

The application of AI to due diligence is nothing new. Correlation Ventures, EQT Ventures and Signalfire are among the companies that currently use algorithms to report their investments. Gartner predicts that 75% of venture capitalists will use AI to make investment decisions in 2025, compared to less than the current 5%. But while some see the value of this technology, the dangers lie beneath the surface. In 2020, the Harvard Business Review (HBR) found that an investment algorithm outperformed novice investors, but it had biases, such as frequently selecting white and male entrepreneurs. HBR noted that this reflects the real world, pointing to the tendency of AI to amplify existing biases.

In more encouraging news, MIT scientists, along with researchers from Cornell and Microsoft, claim to have developed a computer vision algorithm, STEGO, that can identify images down to the pixel. While this may seem trivial, it is a great improvement over the traditional method of “teaching” an algorithm to detect and classify objects in photos and videos.

Traditionally, computer vision algorithms learn to recognize objects (e.g., trees, cars, tumors, etc.) by showing them many examples of objects that have been tagged by humans. STEGO eliminates this tedious and time-consuming process by applying a class tag to each pixel in the image.. The system isn’t perfect (it sometimes confuses oatmeal with pasta, for example), but STEGO manages to segment objects such as roads, people, and traffic signs, researchers say.

When it comes to object recognition, we seem to be approaching the day when academic papers like DALL-E 2, OpenAI’s imaging system, will become products. New research from Columbia University shows a system called Opal, designed to create featured images for news articles from text descriptions, guiding users through the process through visual cues.

When tested with a group of users, the researchers said that those who tested Opal were ” more effective “ to create featured images for articles, creating more than twice as many “usable” results as users who haven’t tried them. It’s not hard to imagine that a tool like Opal can be integrated into content management systems like WordPress, perhaps as a complement or extension.

“Given the text of the article, Opal guides users through a structured search for visual concepts and provides channels for users to illustrate them based on tone, theme, and style. planned illustration of the article “, write the co-authors. ” [Opal] generates various sets of editorial illustrations, graphic resources, and conceptual ideas. »





Leave a Comment