Fake faces created by artificial intelligence (AI) are considered more trustworthy than images of real people, a new study has found. The results highlight the need for safeguards to prevent deep fakes, which have already been used for revenge porn, fraud and propaganda, the researchers behind the report say. The study - by Dr Sophie Nightingale from Lancaster University in the UK and Professor Hany Farid from the University of California, Berkeley, in the US - asked participants to identify a selection of 800 faces as real or fake, and to rate their trustworthiness. After three separate experiments, the researchers found the AI-created synthetic faces were on average rated 7.7% more trustworthy than the average rating for real faces. This is "statistically significant", they add.

For years, beer drinkers have had to pretend to enjoy dodgy-tasting beer served up by hipster breweries and enthusiastic home-brewers. Now researchers in Japan say two different facial expressions can truly reveal whether or not we enjoyed a beer immediately after trying it. In experiments, the scientists used facial recognition technology to scan people's facial expressions to reveal their true beer preferences. 'Lip suck', where the lips are drawn inwards as if we're saying'mmmmm', indicate that we enjoy a beverage, the experts claim. Conversely, 'lip press', where the lips are pressed down on top of each other, reveals that we actually thought a beer tasted horrible.

John Herbst expresses offense at American reluctance to provide MiGs to Ukraine on'Your World.' This is a rush transcript from "Your World," March 18, 2022. This copy may not be in its final form and may be updated. NEIL CAVUTO, FOX NEWS ANCHOR: All right, Vladimir Putin defending his invasion of Ukraine and maybe wincing at all the global notoriety the Ukrainian president is getting, when he never leaves Ukraine, talking to one major legislative body after another of the greatest powers on Earth, as Vladimir Putin tries to explain to a packed crowd in a Moscow stadium that he means no harm, that he is doing good, that he is fighting the good fight, even as that good fight is turning awfully deadly and getting awfully close to a NATO country. In Lviv today, in the western part of the country, a mere 40 miles from the Polish border, the missiles were flying and people were dying. MIKE TOBIN, FOX NEWS CORRESPONDENT: Well, Neil, as you mentioned, for the first time, the -- first time in several days, the relative peace of the western part of the country was shattered, as cruise missiles rained down here perilously close to NATO's eastern flank. What they were after was the Lviv state aircraft repair. What that facility does is customize MiG-29s, so they can be used by the Ukrainian air force. Maxim Kozytskyy, the regional administrator of Lviv says the airstrikes were launched from long-range bombers over the Black Sea. Six of the missiles were launched. Four of them got through. Two of them were intercepted by Ukrainian air defenses. The Ukrainian air force says one of the reasons the cruisers was way able to get through is because they flew so low. They are the Russian X-55s, with a price tag of about a million apiece. South of here, the town of Mariupol, the situation is quite desperate. You know that theater that was being used as a bomb shelter took a direct hit from a Russian aircraft.

Over 100 Guyanese students will benefit from the Spark Programme – an artificial intelligence and leadership initiative – aimed at equipping them to grow their technological skills and create economic opportunities. The programme is in collaboration with two overseas-based Guyanese, Professor and Scientist at the University of Michigan, Jason Mars and Denise Hilliman, a former science educator and the Chief Executive Officer of Lead Mindset – leadership curriculum. The programme is being facilitated by the Ministry of Education. Mars and Hilliman will be sharing their knowledge of artificial intelligence, technology and leadership with students here. "We have come together to do something for our people and to bring the successes we have in the diaspora and come back home to spark the pathway to ignite innovation and perhaps a transformation in technology and economic prosperity by working on what is on the minds of our young people," Mars said at the launch of the programme at the National Centre for Educational Resource Development (NCERD), Kingston, Georgetown.

The COVID-19 pandemic has demonstrated the increasing need of policymakers for timely estimates of macroeconomic variables. A prior UNCTAD research paper examined the suitability of long short-term memory artificial neural networks (LSTM) for performing economic nowcasting of this nature. Here, the LSTM's performance during the COVID-19 pandemic is compared and contrasted with that of the dynamic factor model (DFM), a commonly used methodology in the field. Three separate variables, global merchandise export values and volumes and global services exports, were nowcast with actual data vintages and performance evaluated for the second, third, and fourth quarters of 2020 and the first and second quarters of 2021. In terms of both mean absolute error and root mean square error, the LSTM obtained better performance in two-thirds of variable/quarter combinations, as well as displayed more gradual forecast evolutions with more consistent narratives and smaller revisions. Additionally, a methodology to introduce interpretability to LSTMs is introduced and made available in the accompanying nowcast_lstm Python library, which is now also available in R, MATLAB, and Julia.

One of the most classical results in high-dimensional learning theory provides a closed-form expression for the generalisation error of binary classification with the single-layer teacher-student perceptron on i.i.d. Gaussian inputs. Both Bayes-optimal estimation and empirical risk minimisation (ERM) were extensively analysed for this setting. At the same time, a considerable part of modern machine learning practice concerns multi-class classification. Yet, an analogous analysis for the corresponding multi-class teacher-student perceptron was missing. In this manuscript we fill this gap by deriving and evaluating asymptotic expressions for both the Bayes-optimal and ERM generalisation errors in the high-dimensional regime. For Gaussian teacher weights, we investigate the performance of ERM with both cross-entropy and square losses, and explore the role of ridge regularisation in approaching Bayes-optimality. In particular, we observe that regularised cross-entropy minimisation yields close-to-optimal accuracy. Instead, for a binary teacher we show that a first-order phase transition arises in the Bayes-optimal performance.

Within the Latin American and Caribbean region, it has been recorded that at least 25% of the population lives in informal settlements. Given that their expansion is one of the major problems afflicting these cities, a project is presented, supported by the IDB, which proposes how new technologies are capable of contributing to the identification and detection of these areas in order to intervene in them and help reduce urban informality. Informal settlements, also known as slums, shantytowns, camps or favelas, depending on the country in question, are uncontrolled settlements on land where, in many cases, the conditions for a dignified life are not in place. Through self-built dwellings, these sites are generally the result of the continuous growth of the housing deficit. For decades, the possibility of collecting information about the Earth's surface through satellite imagery has been contributing to the analysis and production of increasingly accurate and useful maps for urban planning.

Understandably, society views Urban Air Mobility (UAM) in this way, however, the rapid increase in use-cases, investments, and infrastructure being introduced around the globe may prove that this assumption may be wrong. In February, the first airport designated to'flying taxis' started being built in Coventry, UK, as a carbon-neutral, zero-emissions hotspot for travel. The vision is that air-taxis or electric vertical take-off and landing (eVTOL) vehicles will provide a platform to transport both people and goods in urban airspace, easing the congestion that we're so used to seeing on the ground. Moreover, the UK Research & Innovation (UKRI) is investing £125 million into its'Future flight challenge' scheme to develop more sustainable methods of flying. São Paulo, Brazil, a city with a heavily-congested road network and host to the highest concentration of helicopters in the world, is working with the Civil Aviation Agency of Brazil (ANAC) to drive forward the type certificate validation process for eVTOL, and Volocopter, the German aircraft manufacturer, is to provide eVTOL's to France and Singapore in the next three years.

Much has been said about the potential of artificial intelligence (AI) to transform how we live, work, and interact with each other. But we must also draw attention to a less discussed, but equally important, question -- do we have the skills required to develop AI inclusively and use it responsibly? AI adoption is accelerating, and the overall market is expected to be worth $190 billion by 2025. By 2030, AI technology will add $15.7 trillion to global gross domestic product (GDP). AI is everywhere -- whether we're aware of it or not.

Pre-trained language models derive substantial linguistic and factual knowledge from the massive corpora on which they are trained, and prompt engineering seeks to align these models to specific tasks. Unfortunately, existing prompt engineering methods require significant amounts of labeled data, access to model parameters, or both. We introduce a new method for selecting prompt templates without labeled examples and without direct access to the model. Specifically, over a set of candidate templates, we choose the template that maximizes the mutual information between the input and the corresponding model output. Figure 1: Performance of template selected by our maximum Across 8 datasets representing 7 distinct NLP mutual information method (MI) compared to tasks, we show that when a template has high the the worst, mean, median, and best prompt on GPT-3 mutual information, it also has high accuracy Davinci (175B). Our method performs at almost oracle on the task. On the largest model, selecting levels, without labels or access to model weights.