Large language models can encode a wealth of semantic knowledge about the world. Such knowledge could be extremely useful to robots aiming to act upon high-level, temporally extended instructions expressed in natural language. However, a significant weakness of language models is that they lack real-world experience, which makes it difficult to leverage them for decision making within a given embodiment. For example, asking a language model to describe how to clean a spill might result in a reasonable narrative, but it may not be applicable to a particular agent, such as a robot, that needs to perform this task in a particular environment. We propose to provide real-world grounding by means of pretrained skills, which are used to constrain the model to propose natural language actions that are both feasible and contextually appropriate. The robot can act as the language model's "hands and eyes," while the language model supplies high-level semantic knowledge about the task. We show how low-level skills can be combined with large language models so that the language model provides high-level knowledge about the procedures for performing complex and temporally extended instructions, while value functions associated with these skills provide the grounding necessary to connect this knowledge to a particular physical environment. We evaluate our method on a number of real-world robotic tasks, where we show the need for real-world grounding and that this approach is capable of completing long-horizon, abstract, natural language instructions on a mobile manipulator. The project's website and the video can be found at say-can.github.io. (a) Large Language Models (LLMs) (b) SayCan

No abstract

No abstract

Stuart Bretschneider is associate dean and chair of the Department of Public Administration, Maxwell School of Citizenship and Public Affairs, Syracuse University. He also holds one of the University's Laura J. and L. Douglas Meredith Professorships for Teaching Excellence. His research focuses on how public organizations make use of information technology and the effects of those technologies on public organizations; how public organizations employ forecasting technology and organize to carry out forecasting activities; and how sector differences affect administrative processes. HeTh is case study reports an innovative e-government experiment by a local government in Seoul, South Korea-Gangnam-gu. A new local political leadership in Gangnam made strategic use of e-government applications to exert greater political control over the local civil service bureaucracy. Th e authors fi nd that e-government applications possess political properties that can be applied eff ectively by the political leadership as instruments to improve control over the government bureaucracy as well as to enhance essential government accountability and transparency. Th e political circumstances underlying e-government development as well as its impact on local government are reported, along with key variables associated with this innovation and directions for future research.

E-government has been touted by many as a technological answer to improve citizen participation, government accountability, and transparency by facilitating a greater level of communication and flow of public information between citizens and the government. This article examines how political environment, government structure, and the nature of individual e-government applications influence the likelihood of adoption. Using data obtained from multiple sources, logistic regressions are conducted on a sample of six e-government applications that possess varying degrees of communicative and organizational impacts on the government to observe how different factors influence their adoption. Findings include a general disinclination for adopting e-government applications with high communicative impact; however, such disinclination dissipated when there was a high level of political competition in the area and perceived demand for online communication; active traditional channels of political communication, such as political parties and accessibility to local council members, reduced the likelihood of adoption; the preferences of the elected mayors coincided with the perceptions of nonelected officials who favor e-government applications that would reduce the workload while disfavoring applications that would increase it.

Twenty-eight questionnaires were gathered (54.9% response rate), including 20 from Southampton and eight from BSMS. Long-term knowledge retention and better understanding of the material were rated 8.1 and 7.9 out of 10, respectively. Eight responses were from currently practising doctors, who rated how much they now use their teaching skills as doctors as 8.9 out of 10. Of the eight doctors, seven gained points for their foundation programme applications as a direct result of near-peer teaching. The most common motivator for engaging in teaching was to improve subject matter knowledge and the most common benefit was improved communication skills. There are numerous advantages to being a near-peer teacher in medical school DISCUSSION: There are numerous advantages to being a near-peer teacher in medical school, which include knowledge improvement, transferrable professional skills and employability. These initial results support the hypothesised benefits to the teachers and provide a foundation for further longitudinal studies.

Reinforcement learning provides a general framework for learning robotic skills while minimizing engineering effort. However, most reinforcement learning algorithms assume that a well-designed reward function is provided, and learn a single behavior for that single reward function. Such reward functions can be difficult to design in practice. Can we instead develop efficient reinforcement learning methods that acquire diverse skills without any reward function, and then re-purpose these skills for downstream tasks? In this paper, we demonstrate that a recently proposed unsupervised skill discovery algorithm can be extended into an efficient off-policy method, making it suitable for performing unsupervised reinforcement learning in the real world. Firstly, we show that our proposed algorithm provides substantial improvement in learning efficiency, making rewardfree real-world training feasible. Secondly, we move beyond the simulation environments and evaluate the algorithm on real physical hardware. On quadrupeds, we observe that locomotion skills with diverse gaits and different orientations emerge without any rewards or demonstrations. We also demonstrate that the learned skills can be composed using model predictive control for goal-oriented navigation, without any additional training.

This study aims to address strategies, models, and the motivation behind smart cities by analyzing two smart city project cases in medium-sized cities, i.e., Gimpo and Namyangju in South Korea. The case of Smartopia Gimpo represents a top-down, infrastructure-focused smart city innovation that invested in building state-of-the-art big data infrastructure for crime prevention, traffic alleviation, environmental preservation, and disaster management. On the other hand, Namyangju 4.0 represents a strategy focused on internal process innovation through extensive employee training and education regarding smart city concepts and emphasizing data-driven (rather than infrastructure-driven) policy decision making. This study explores two smart city strategies and how they resulted in distinctively different outcomes. We found that instilling a culture of innovation through the training of government managers and frontline workers is a critical component in achieving a holistic and sustainable smart city transformation that can survive leadership changes.