In this paper, we study how well humans can detect text generated by commercial LLMs (GPT-4o, Claude, o1). We hire annotators to read 300 non-fiction English articles, label them as either human-written or AI-generated, and provide paragraph-length explanations for their decisions. Our experiments show that annotators who frequently use LLMs for writing tasks excel at detecting AI-generated text, even without any specialized training or feedback. In fact, the majority vote among five such "expert" annotators misclassifies only 1 of 300 articles, significantly outperforming most commercial and open-source detectors we evaluated even in the presence of evasion tactics like paraphrasing and humanization. Qualitative analysis of the experts' free-form explanations shows that while they rely heavily on specific lexical clues ('AI vocabulary'), they also pick up on more complex phenomena within the text (e.g., formality, originality, clarity) that are challenging to assess for automatic detectors. We release our annotated dataset and code to spur future research into both human and automated detection of AI-generated text.

AI humanizers are a new class of online software tools meant to paraphrase and rewrite AI-generated text in a way that allows them to evade AI detection software. We study 19 AI humanizer and paraphrasing tools and qualitatively assess their effects and faithfulness in preserving the meaning of the original text. We show that many existing AI detectors fail to detect humanized text. Finally, we demonstrate a robust model that can detect humanized AI text while maintaining a low false positive rate using a data-centric augmentation approach. We attack our own detector, training our own fine-tuned model optimized against our detector's predictions, and show that our detector's cross-humanizer generalization is sufficient to remain robust to this attack.

Antibody therapies have been employed to address some of today's most challenging diseases, but must meet many criteria during drug development before reaching a patient. Humanization is a sequence optimization strategy that addresses one critical risk called immunogenicity -- a patient's immune response to the drug -- by making an antibody more'human-like' in the absence of a predictive lab-based test for immunogenicity. However, existing humanization strategies generally yield very few humanized candidates, which may have degraded biophysical properties or decreased drug efficacy. Here, we re-frame humanization as a conditional generative modeling task, where humanizing mutations are sampled from a language model trained on human antibody data. We describe a sampling process that incorporates models of therapeutic attributes, such as antigen binding affinity, to obtain candidate sequences that have both reduced immunogenicity risk and maintained or improved therapeutic properties, allowing this algorithm to be readily embedded into an iterative antibody optimization campaign. We demonstrate in silico and in lab validation that in real therapeutic programs our generative humanization method produces diverse sets of antibodies that are both (1) highly-human and (2) have favorable therapeutic properties, such as improved binding to target antigens. Antibodies are the fastest growing drug class, with approved molecules treating a breadth of disorders ranging from cancer to autoimmune disease to infectious disease (Carter & Lazar, 2018). Many candidate therapeutic antibodies are derived from non-human e.g., murine or camelid sources, and modern antibody formats such as multi-specifics or antibody-drug conjugates can require heavy sequence engineering after discovery. This increases the risk of immunogenicity, where Anti-Drug Antibodies (ADAs) result in either fast clearance of the drug or adverse events (Hwang & Foote, 2005). While antibody sequence humanness is only roughly correlated with immunogenicity, humanization is widely employed to decrease immunogenicity risk (Prihoda et al., 2022).

New research that I have conducted with my colleagues at the University of Oxford--Felipe Thomaz, Rhonda Hadi and Andrew Stephen--reveals that making chatbots more humanlike is a double-edged sword. On one hand, when customers are neutral, happy or even sad, interacting with humanized chatbots can boost customer satisfaction. Yet, when customers are angry, interacting with humanized chatbots only increases their dissatisfaction, meaning that a company's most unsatisfied customers are often handled the most poorly. More important, this lower satisfaction doesn't just affect the single chat interaction or the customer's feelings about the chatbot itself; it extends to negative feelings toward the entire company and decreases consumers' desire to purchase from that company in the future. Chatbots are becoming more common across a host of industries, as companies replace human customer-service agents on their websites, social-media pages and messaging services.

If you want to imagine the future of healthcare, you can do no better than to read cardiologist and bestselling author Eric Topol's trilogy on the subject: "The Creative Destruction of Medicine," "The Patient Will See You Now," and "Deep Medicine." "Deep Medicine" bears a paradoxical subtitle: "How Artificial Intelligence Can Make Healthcare Human Again." The book describes the growing interaction of human and machine brains. Topol envisions a symbiosis, with people and machines working together to assist patients in ways that neither can do alone. In the process, healthcare providers will shed some of the mind-numbing rote tasks they endure today, giving them more time to focus on patients.

Intelligent algorithms are used to create paintings, write poems, and compose music. According to a study by an international team of researchers from the Massachusetts Institute of Technology (MIT), and the Center of Humans and Machines at the Max Planck Institute for Human Development, whether people perceive artificial intelligence (AI) as the ingenious creator of art or simply another tool used by artists depends on how information about AI art is presented. The results were published in the journal iScience. In October 2018, a work of art by Edmond de Belamie, which was created with the help of an intelligent algorithm, was auctioned for 432,500 USD at Christie's Auction House. According to Christie's auction advertisement, the portrait was created by artificial intelligence (AI).

AI is promising the automation of customer engagement to scale intelligent, real-time CX to survive digital Darwinism. The future has two possibilities however. We can automate and scale the standards and capabilities for CX as we've come to define and shape them over the decades. Or, we can use AI as a mechanism to deliver personalized, on-demand experiences that connected customers are learning to expect from their favorite digital-native apps and services. In this special session, leading digital anthropologist, best-selling author and keynote speaker Brian Solis will share how AI and a little bit of ingenuity and imagination can unlock an entirely new generation of customer experiences that become the standard for everyone else to follow.

We explore models for translating abstract musical ideas (scores, rhythms) into expressive performances using Seq2Seq and recurrent Variational Information Bottleneck (VIB) models. Though Seq2Seq models usually require painstakingly aligned corpora, we show that it is possible to adapt an approach from the Generative Adversarial Network (GAN) literature (e.g. Pix2Pix (Isola et al., 2017) and Vid2Vid (Wang et al. 2018a)) to sequences, creating large volumes of paired data by performing simple transformations and training generative models to plausibly invert these transformations. Music, and drumming in particular, provides a strong test case for this approach because many common transformations (quantization, removing voices) have clear semantics, and models for learning to invert them have real-world applications. Focusing on the case of drum set players, we create and release a new dataset for this purpose, containing over 13 hours of recordings by professional drummers aligned with fine-grained timing and dynamics information. We also explore some of the creative potential of these models, including demonstrating improvements on state-of-the-art methods for Humanization (instantiating a performance from a musical score).

Think of artificial intelligence (AI) and you'll think of factories of robots doing manufacturing jobs. You may think of logistics and evoke imagery of robots doing their thing for the likes of Amazon. However, whilst the media often focuses on these radical and wide-reaching applications of AI, there are other forms of adoption which are quietly stirring a revolution. We see this in the way AI is used, and being developed, within the insurance sector. What's particularly interesting is that these wide-reaching forms of AI aren't the headline makers.

HALF MOON BAY, Calif.--Artificial intelligence is routinely touted at tech conferences and elsewhere as the "Next Big Thing" that is going to transform the customer experience and the ability of companies to better sell and market their wares. But there were also skeptical and cautionary notes sounded here, even from vendors, at the Connected Enterprise conference (running Oct. 22-25) sponsored by Constellation Research. "There are a lot of misconceptions about what AI can do in the enterprise. I would focus on really picking a specific problem," said Inhi Cho Suh, general manager of Watson Customer Engagement at IBM. For customers of IBM's Watson AI supercomputer services, Suh said it's important to focus on precise algorithms for small sets of data.