This open access book provides a comprehensive overview of the state of the art in research and applications of Foundation Models and is intended for readers familiar with basic Natural Language Processing (NLP) concepts. Over the recent years, a revolutionary new paradigm has been developed for training models for NLP. These models are first pre-trained on large collections of text documents to acquire general syntactic knowledge and semantic information. Then, they are fine-tuned for specific tasks, which they can often solve with superhuman accuracy. When the models are large enough, they can be instructed by prompts to solve new tasks without any fine-tuning. Moreover, they can be applied to a wide range of different media and problem domains, ranging from image and video processing to robot control learning. Because they provide a blueprint for solving many tasks in artificial intelligence, they have been called Foundation Models. After a brief introduction to basic NLP models the main pre-trained language models BERT, GPT and sequence-to-sequence transformer are described, as well as the concepts of self-attention and context-sensitive embedding. Then, different approaches to improving these models are discussed, such as expanding the pre-training criteria, increasing the length of input texts, or including extra knowledge. An overview of the best-performing models for about twenty application areas is then presented, e.g., question answering, translation, story generation, dialog systems, generating images from text, etc. For each application area, the strengths and weaknesses of current models are discussed, and an outlook on further developments is given. In addition, links are provided to freely available program code. A concluding chapter summarizes the economic opportunities, mitigation of risks, and potential developments of AI.

We make a connection between multicalibration and property elicitation and show that (under mild technical conditions) it is possible to produce a multicalibrated predictor for a continuous scalar distributional property $\Gamma$ if and only if $\Gamma$ is elicitable. On the negative side, we show that for non-elicitable continuous properties there exist simple data distributions on which even the true distributional predictor is not calibrated. On the positive side, for elicitable $\Gamma$, we give simple canonical algorithms for the batch and the online adversarial setting, that learn a $\Gamma$-multicalibrated predictor. This generalizes past work on multicalibrated means and quantiles, and in fact strengthens existing online quantile multicalibration results. To further counter-weigh our negative result, we show that if a property $\Gamma^1$ is not elicitable by itself, but is elicitable conditionally on another elicitable property $\Gamma^0$, then there is a canonical algorithm that jointly multicalibrates $\Gamma^1$ and $\Gamma^0$; this generalizes past work on mean-moment multicalibration. Finally, as applications of our theory, we provide novel algorithmic and impossibility results for fair (multicalibrated) risk assessment.

State-of-the-art abstractive summarization systems frequently hallucinate content that is not supported by the source document, mainly due to noise in the training dataset. Existing methods opt to drop the noisy samples or tokens from the training set entirely, reducing the effective training set size and creating an artificial propensity to copy words from the source. In this work, we propose a training objective for abstractive summarization based on rejection learning, in which the model learns whether or not to reject potentially noisy tokens. We further propose a regularized decoding objective that penalizes non-factual candidate summaries during inference by using the rejection probability learned during training. We show that our method considerably improves the factuality of generated summaries in automatic and human evaluations when compared to five baseline models and that it does so while increasing the abstractiveness of the generated summaries.

The World Customs Organization (WCO) recently conducted a BACUDA Data Analytics workshop for the Maldives Customs Service with 41 participants from the 30th of January to the 1st of February in Male, Maldives. The mission was financed by the Customs Cooperation Fund of Korea (CCF-Korea) and took place under the WCO's BACUDA initiative, the WCO capacity building project on Data Analytics. WCO experts and two BACUDA Scholarship graduates led the workshop. They delivered various sessions to equip the customs officials with the latest data analytics tools and techniques. One of the key highlights of the workshop was a hands-on session where the participants learned how to use Python language to work with the AI HS algorithm developed through the BACUDA project.

Couples for whom the spark may have gone from their relationship will find it a little easier to rekindle the romance this Valentine's Day. Moonpig, the online customised greetings card retailer, is trialling using ChatGPT to generate personalised messages or poems for loved ones. ChatGPT is a generative artificial intelligence (AI) tool, developed by San Francisco company OpenAI, in which Microsoft recently invested billions of dollars. In the few months since a beta version of ChatGPT was released to the world, it has rapidly become an integral part of many people's lives. Estate agents in the United States now say they can't live without the tool's automation to write up property descriptions.

The following MBW op/ed comes from Michael Nash (pictured inset, below), Executive Vice President and Chief Digital Officer, Universal Music Group. AI is transforming the ways we live, work and play – from chatbots that answer complex questions to systems that can write passable screenplays to programs that have passed part of a bar exam in the US. AI is now creating imagery comparable to professional artists -- with one AI-generated portrait being sold for £40,000 at Sotheby's and another composition winning a State Fair competition in Colorado. Learning from millions of images with associated descriptions of subject matter, composition, methodology and other inputs, the most advanced AI can now generate derivative output that closely mimics original creators' distinct styles. In some cases, this is used to produce outright fakes.

NEW DELHI: While artificial intelligence is being used in all fields, its application in the healthcare sector is vital but there are several legal issues involved, as there are no specific laws to deal with it and the question of accountability for errors in technology remain. This was widely felt during the concluding session of 10th international conference on transforming healthcare with IT organized by the Apollo hospitals and Apollo telemedicine networking foundation. While speaking on the legal implication in use of AI in clinical practice: gray areas, Bagmishika Puhan, associate partner, TMT law practice said that currently there are no well defined regulations in place to address the legal and ethical issues that may arise due to the use of AI in healthcare sector, which is required. "Who should be responsible for errors in a medical device, diagnosis and treatment enabled by AI," she questioned, adding that adoption of AI should steer away from flawed algorithms, human bias and any potential discrimination, exclusions. She further said that the patient must be made aware of technologies deployed as much as they need to be aware of the consequences because in digital services, lack of physical presence creates an impression of vulnerability in the minds of the patients.

Artificial intelligence (AI) is both omnipresent and conceptually slippery, making it notoriously hard to regulate. Fortunately for the rest of the world, two major experiments in the design of AI governance are currently playing out in Europe and China. The European Union (EU) is racing to pass its draft Artificial Intelligence Act, a sweeping piece of legislation intended to govern nearly all uses of AI. Meanwhile, China is rolling out a series of regulations targeting specific types of algorithms and AI capabilities. For the host of countries starting their own AI governance initiatives, learning from the successes and failures of these two initial efforts to govern AI will be crucial.

The rapid adoption of facial recognition (FR) technology by both government and commercial entities in recent years has raised concerns about civil liberties and privacy. In response, a broad suite of so-called "anti-facial recognition" (AFR) tools has been developed to help users avoid unwanted facial recognition. The set of AFR tools proposed in the last few years is wide-ranging and rapidly evolving, necessitating a step back to consider the broader design space of AFR systems and long-term challenges. This paper aims to fill that gap and provides the first comprehensive analysis of the AFR research landscape. Using the operational stages of FR systems as a starting point, we create a systematic framework for analyzing the benefits and tradeoffs of different AFR approaches. We then consider both technical and social challenges facing AFR tools and propose directions for future research in this field.

In a context in which art competitions are beginning to be won by AI systems, artists must defend themselves, at least to identify what a machine has done from what a human has done. Now it's the University of Chicago that aims to help with a tool called Glaze, to prevent artificial intelligence (AI) models from learning an artist's style, in response to the threat that AI-generated images will win art contests. To prevent the AI from learning an artist's style, the artist can upload a digital version of their work to Glaze and choose a different art type than their own. The tool will make pixel-level changes that the AI would associate with another style. The University of Chicago team acknowledges that the tool does not guarantee absolute protection, but its goal is to fill the gap until laws, regulations, and policies are updated.