Artificial intelligence (AI) will change the way we learn and work in the near future. Nearly 400 million workers globally will change their occupations in the next 10 years, and business schools are uniquely situated to respond to the shifts coming to the future of work. However, a recent study, "Implications of Artificial Intelligence on Business Schools and Lifelong Learning," shows that business schools remain cautious in adapting management education to address the changing needs of students, workers and organizations, writes Anne Trumbore in this opinion piece. Trumbore, one of the study's coauthors, is senior director of Wharton Online, a strategic digital learning initiative at the Wharton School of the University of Pennsylvania. In the past few weeks, COVID 19 has moved hundreds of millions of students around the globe from physical to online classes.

There is a very high demand for AI & ML professionals who are qualified enough to do state-of-the-art research and engineering. At the same time, the supply of specialized AI talent is scarce – though the situation is gradually improving thanks to the new Master's and Ph.D. programs in data science and machine learning that have been launched all over the world in the last few years. Still, hiring a good ML engineer remains a challenging task for recruiters – not only because of the scarcity of AI talent but also due to a lack of relevant experience among recruiting specialists. Artificial Intelligence remains a new and obscure field for most recruiters. In this article, we share with you our detailed guidelines for recruiting AI & ML professionals, including the skills to look for, the recruiting strategies to apply depending on the situation, and the advantages you can show off to attract the top talent.

Of all the cool things about octopuses (and there are a lot), their arms may rank among the coolest. Two-thirds of an octopus's neurons are in its arms, meaning each arm literally has a mind of its own. The hundreds of suckers that cover their arms can form strong seals even on rough surfaces underwater. Imagine if a robot could do all that. Researchers at Harvard's Wyss Institute for Biologically Inspired Engineering and John A. Paulson School of Engineering and Applied Sciences (SEAS) and colleagues from Beihang University have developed an octopus-inspired soft robotic arm that can grip, move, and manipulate a wide range of objects.

Learning from non-stationary data remains a great challenge for machine learning. Continual learning addresses this problem in scenarios where the learning agent faces a stream of changing tasks. In these scenarios, the agent is expected to retain its highest performance on previous tasks without revisiting them while adapting well to the new tasks. Two new recent continual-learning scenarios have been proposed. In meta-continual learning, the model is pre-trained to minimize catastrophic forgetting when trained on a sequence of tasks. In continual-meta learning, the goal is faster remembering, i.e., focusing on how quickly the agent recovers performance rather than measuring the agent's performance without any adaptation. Both scenarios have the potential to propel the field forward. Yet in their original formulations, they each have limitations. As a remedy, we propose a more general scenario where an agent must quickly solve (new) out-of-distribution tasks, while also requiring fast remembering. We show that current continual learning, meta learning, meta-continual learning, and continual-meta learning techniques fail in this new scenario. Accordingly, we propose a strong baseline: Continual-MAML, an online extension of the popular MAML algorithm. In our empirical experiments, we show that our method is better suited to the new scenario than the methodologies mentioned above, as well as standard continual learning and meta learning approaches.

Natural beings undergo a morphological development process of their bodies while they are learning and adapting to the environments they face from infancy to adulthood. In fact, this is the period where the most important learning pro-cesses, those that will support learning as adults, will take place. However, in artificial systems, this interaction between morphological development and learning, and its possible advantages, have seldom been considered. In this line, this paper seeks to provide some insights into how morphological development can be harnessed in order to facilitate learning in em-bodied systems facing tasks or domains that are hard to learn. In particular, here we will concentrate on whether morphological development can really provide any advantage when learning complex tasks and whether its relevance towards learning in-creases as tasks become harder. To this end, we present the results of some initial experiments on the application of morpho-logical development to learning to walk in three cases, that of a quadruped, a hexapod and that of an octopod. These results seem to confirm that as task learning difficulty increases the application of morphological development to learning becomes more advantageous.

Good prediction is necessary for autonomous robotics to make informed decisions in dynamic environments. Improvements can be made to the performance of a given data-driven prediction model by using better sampling strategies when collecting training data. Active learning approaches to optimal sampling have been combined with the mathematically general approaches to incentivizing exploration presented in the curiosity literature via model-based formulations of curiosity. We present an adversarial curiosity method which maximizes a score given by a discriminator network. This score gives a measure of prediction certainty enabling our approach to sample sequences of observations and actions which result in outcomes considered the least realistic by the discriminator. We demonstrate the ability of our active sampling method to achieve higher prediction performance and higher sample efficiency in a domain transfer problem for robotic manipulation tasks. We also present a validation dataset of action-conditioned video of robotic manipulation tasks on which we test the prediction performance of our trained models.

We introduce a new way of learning to encode position information for non-recurrent models, such as Transformer models. Unlike RNN and LSTM, which contain inductive bias by loading the input tokens sequentially, non-recurrent models are less sensitive to position. The main reason is that position information among input units is not inherently encoded, i.e., the models are permutation equivalent; this problem justifies why all of the existing models are accompanied by a sinusoidal encoding/embedding layer at the input. However, this solution has clear limitations: the sinusoidal encoding is not flexible enough as it is manually designed and does not contain any learnable parameters, whereas the position embedding restricts the maximum length of input sequences. It is thus desirable to design a new position layer that contains learnable parameters to adjust to different datasets and different architectures. At the same time, we would also like the encodings to extrapolate in accordance with the variable length of inputs. In our proposed solution, we borrow from the recent Neural ODE approach, which may be viewed as a versatile continuous version of a ResNet. This model is capable of modeling many kinds of dynamical systems. We model the evolution of encoded results along position index by such a dynamical system, thereby overcoming the above limitations of existing methods. We evaluate our new position layers on a variety of neural machine translation and language understanding tasks, the experimental results show consistent improvements over the baselines.

While control architectures and other information processing approaches are certainly important for robotic intelligence, our understanding of how this intelligence comes about has expanded in the last decades to include the morphology of the robot and its environment, as well as their mutual interactions [1]-[3], The field of Artificial Embodied Intelligence (AEI) [4], [5], which postulates robot intelligence as the result of the interaction between brain, morphology and the environment the robot must operate in, is growing. Currently, this view on the emergence of intelligence has expanded and now it tries to address the fact that intelligent systems must be able to operate in sequences of environments that are generally unknown at design time [6]. In other words, we are facing open-ended learning problems and, by definition, these types of problems imply that robots cannot be completely defined at design time, as, at that time, we do not know what skills the robot will require in order to achieve its purpose. In fact, not even the goals that need achieving are known. Developmental Robotics (DR) [7] is one of the approaches proposed to try to address these issues.

Following the release of the 2020 State of Enterprise Machine Learning report, we created an interactive data visualization so anyone can explore the survey data, conduct analysis, and see how a company's machine learning efforts compare to others like it. The State of Enterprise Machine Learning (ML) experience shares eight questions that were posed in our survey and the associated results. After exploring the data, download the full report to read our assessments and predictions about where ML development is headed. Our report shares findings from nearly 750 survey respondents whom we polled in the fall of 2019. However, if you want to see how other companies of a similar size to yours are using machine learning, the interactive experience allows you to test your own hypotheses and arrive at findings tailored to you.

We hypothesize that curiosity is a mechanism found by evolution that encourages meaningful exploration early in an agent's life in order to expose it to experiences that enable it to obtain high rewards over the course of its lifetime. We formulate the problem of generating curious behavior as one of meta-learning: an outer loop will search over a space of curiosity mechanisms that dynamically adapt the agent's reward signal, and an inner loop will perform standard reinforcement learning using the adapted reward signal. However, current meta-RL methods based on transferring neural network weights have only generalized between very similar tasks. To broaden the generalization, we instead propose to meta-learn algorithms: pieces of code similar to those designed by humans in ML papers. Our rich language of programs combines neural networks with other building blocks such as buffers, nearest-neighbor modules and custom loss functions. We demonstrate the effectiveness of the approach empirically, finding two novel curiosity algorithms that perform on par or better than human-designed published curiosity algorithms in domains as disparate as grid navigation with image inputs, acrobot, lunar lander, ant and hopper.