The ability to base current computations on memories from the past is critical for many cognitive tasks such as story understanding.

Wehope theevidence from ourpaper motivates machine learning researchers to pay closer attention to how their training loss affects the modelbehaviortheyactuallywant.

Microsoft has shared a video in which David Weston, who holds the title of Corporate Vice President of Enterprise & Security, shares his vision of how Windows will work in 2030. Here's what his vision looks like: As it turns out, the use of AI agents will mean that we'll no longer need keyboards and mice for our computers. According to Weston, keyboards and mice will feel "as alien as DOS" to Gen Z (which seems an odd statement given that Gen Z is already between 13 and 30 years old). Judging by the comments on the video, few are interested in the future being painted. For example, many users say they would rather switch to Linux or Mac than run Windows without a keyboard and mouse.

Microsoft has released a white paper of sorts outlining what the company is doing to secure user data within Windows Recall, the controversial Windows feature that takes snapshots of your activity for later searching. As of late last night, Microsoft still hasn't said whether they will release Recall to the Windows Insider channels for further testing as originally planned. In fact, Microsoft's paper says very little about Recall as a product or when they will push Recall live to the public. Recall was first launched back in May as part of the Windows 11 24H2 update and it uses the local AI capabilities of Copilot PCs. The idea is that Recall captures periodic snapshots of your screen, then uses optical character recognition plus AI-driven techniques to translate and understand your activity.

In Isaac Asimov's classic science fiction story "Robbie," the Weston family owns a robot who serves as a nursemaid and companion for their precocious preteen daughter, Gloria. Gloria and the robot Robbie are friends; their relationship is affectionate and mutually caring. Gloria regards Robbie as her loyal and dutiful caretaker. However, Mrs. Weston becomes concerned about this "unnatural" relationship between the robot and her child and worries about the possibility of Robbie causing harm to Gloria (despite it's being explicitly programmed to not do so); it is clear she is jealous. After several failed attempts to wean Gloria off Robbie, her father, exasperated and worn down by the mother's protestations, suggests a tour of a robot factory--there, Gloria will be able to see that Robbie is "just" a manufactured robot, not a person, and fall out of love with it.

Exemplar-based generative models for open-domain conversation produce responses based on the exemplars provided by the retriever, taking advantage of generative models and retrieval models. However, they often ignore the retrieved exemplars while generating responses or produce responses over-fitted to the retrieved exemplars. In this paper, we argue that these drawbacks are derived from the one-to-many problem of the open-domain conversation. When the retrieved exemplar is relevant to the given context yet significantly different from the gold response, the exemplar-based generative models are trained to ignore the exemplar since the exemplar is not helpful for generating the gold response. On the other hand, when the retrieved exemplar is lexically similar to the gold response, the generative models are trained to rely on the exemplar highly. Therefore, we propose a training method selecting exemplars that are semantically relevant to the gold response but lexically distanced from the gold response to mitigate the above disadvantages. In the training phase, our proposed training method first uses the gold response instead of dialogue context as a query to select exemplars that are semantically relevant to the gold response. And then, it eliminates the exemplars that lexically resemble the gold responses to alleviate the dependency of the generative models on that exemplars. The remaining exemplars could be irrelevant to the given context since they are searched depending on the gold response. Thus, our proposed training method further utilizes the relevance scores between the given context and the exemplars to penalize the irrelevant exemplars. Extensive experiments demonstrate that our proposed training method alleviates the drawbacks of the existing exemplar-based generative models and significantly improves the performance in terms of appropriateness and informativeness.

Last April, Facebook's AI research lab (FAIR) announced and released as open source its BlenderBot social chat app. While the neophyte AI immediately proved far less prone to racist outbursts than previous attempts, BlenderBot was not without its shortcomings. For one, the system had the recollection capacity of a goldfish -- any subject or data point the AI wasn't initially trained simply didn't exist in its online reality, as evidenced by the OG BB's continued insistence that Tom Brady still plays for the New England Patriots. For another, due to its limited knowledge of current events, the system had a strong tendency to hallucinate knowledge, like a digital Dunning-Kruger effect. But the advancements BlenderBot 2.0 displays, which FAIR debuted on Friday, should make the AI far more sociable, knowledgeable, and capable.

Demand for digital automation skills is rising in the UK as businesses try to find ways to support workers by removing repetitive tasks, according to research. The need for automation proficiency is accelerating faster in non-technical roles such as management consultants and sales directors, up by 125% since 2019, than in traditional IT roles, which are up by 96%, according to a survey carried out by specialist recruiter Robert Half and global labour market trends expert Burning Glass. According to the research, which analysed nine million UK job postings, the number of roles requiring automation capabilities is set to reach around 90,000 this year, compared with 39,323 in 2019. "These findings show that businesses understand they need to find ways to support and supplement their workers by removing more repetitive tasks and freeing them up to focus on more value-add activity," said Matt Weston, managing director at Robert Half UK. Sales directors are increasingly being tasked with improving efficiency, and are hence turning to automation, including software that auto-fills client contact forms during phone calls or using artificial intelligence to support on-the-job training, according to Weston.

Open-domain dialogue agents have vastly improved, but still confidently hallucinate knowledge or express doubt when asked straightforward questions. In this work, we analyze whether state-of-the-art chit-chat models can express metacognition capabilities through their responses: does a verbalized expression of doubt (or confidence) match the likelihood that the model's answer is incorrect (or correct)? We find that these models are poorly calibrated in this sense, yet we show that the representations within the models can be used to accurately predict likelihood of correctness. By incorporating these correctness predictions into the training of a controllable generation model, we obtain a dialogue agent with greatly improved linguistic calibration.

To quantify how well natural language understanding models can capture consistency in a general conversation, we introduce the DialoguE COntradiction DEtection task (DECODE) and a new conversational dataset containing both human-human and human-bot contradictory dialogues. We then compare a structured utterance-based approach of using pre-trained Transformer models for contradiction detection with the typical unstructured approach. Results reveal that: (i) our newly collected dataset is notably more effective at providing supervision for the dialogue contradiction detection task than existing NLI data including those aimed to cover the dialogue domain; (ii) the structured utterance-based approach is more robust and transferable on both analysis and out-of-distribution dialogues than its unstructured counterpart. We also show that our best contradiction detection model correlates well with human judgments and further provide evidence for its usage in both automatically evaluating and improving the consistency of state-of-the-art generative chatbots.