Tesla has settled with a former employee that it sued for downloading data related to its Autopilot feature, Reuters has reported. Tesla filed the lawsuit against Cao Guangzhi back in 2019, accusing its former engineer of copying data to an iCloud account and taking it to his new employer, China's XMotors (owned by Xpeng). Cao reportedly made a monetary payment to Tesla as part of the terms of settlement, but the amount and other details were not disclosed. Cao's legal representative confirmed the settlement, saying he never provided Tesla information to XMotors or any other company. XMotors was not a party in the case, and said it developed its own self-driving technology in-house and respected intellectual property rights.

If you have ever used Google Maps on your phone without fiddling with the location settings, it goes without saying that the tech giant knows everywhere you've been. The really bad news is that even if you have previously tried to stop Google tracking your every movement, the company may have done so anyway. On Friday the Australian Competition and Consumer Commission (ACCC) won a legal action in the federal court, which ruled that, thanks to a peculiar set-up that required a user to check "No" or "Do Not Collect" to both "Location History" and "Web & App Activity" on some Android and Pixel phones, someone who ticked "No" to just one would still end up being tracked. We asked Dr Katharine Kemp, a legal academic from the University of New South Wales whose focus is consumer law, and the Australian cryptographer Vanessa Teague for their thoughts on the significance of the decision and how a person might go about securing their devices. Kemp, an Apple user herself, says that for many consumers, today's decision may not actually mean much, as the decision only related to Android users and Google has since updated the settings that formed the basis of the ACCC's complaint.

Hawkes processes are a popular means of modeling the event times of self-exciting phenomena, such as earthquake strikes or tweets on a topical subject. Classically, these models are fit to historical event time data via likelihood maximization. However, in many scenarios, the exact times of historical events are not recorded for either privacy (e.g., patient admittance to hospitals) or technical limitations (e.g., most transport data records the volume of vehicles passing loop detectors but not the individual times). The interval-censored setting denotes when only the aggregate counts of events at specific time intervals are observed. Fitting the parameters of interval-censored Hawkes processes requires designing new training objectives that do not rely on the exact event times. In this paper, we propose a model to estimate the parameters of a Hawkes process in interval-censored settings. Our model builds upon the existing Hawkes Intensity Process (HIP) of in several important directions. First, we observe that while HIP is formulated in terms of expected intensities, it is more natural to work instead with expected counts; further, one can express the latter as the solution to an integral equation closely related to the defining equation of HIP. Second, we show how a non-homogeneous Poisson approximation to the Hawkes process admits a tractable likelihood in the interval-censored setting; this approximation recovers the original HIP objective as a special case, and allows for the use of a broader class of Bregman divergences as loss function. Third, we explicate how to compute a tighter approximation to the ground truth in the likelihood. Finally, we show how our model can incorporate information about varying interval lengths. Experiments on synthetic and real-world data confirm our HIPPer model outperforms HIP and several other baselines on the task of interval-censored inference.

Large pretrained language models like GPT-3 have acquired a surprising ability to perform zero-shot classification (ZSC). For example, to classify review sentiments, we can "prompt" the language model with the review and the question "Is the review positive?" as the context, and ask it to predict whether the next word is "Yes" or "No". However, these models are not specialized for answering these prompts. To address this weakness, we propose meta-tuning, which trains the model to specialize in answering prompts but still generalize to unseen tasks. To create the training data, we aggregated 43 existing datasets, annotated 441 label descriptions in total, and unified them into the above question answering (QA) format. After meta-tuning, our model outperforms a same-sized QA model for most labels on unseen tasks, and we forecast that the performance would improve for even larger models. Therefore, measuring ZSC performance on non-specialized language models might underestimate their true capability, and community-wide efforts on aggregating datasets and unifying their formats can help build models that understand prompts better.

Over the last decade, we have heard a lot of doom-saying about how artificial intelligence (AI) would result in the loss of huge numbers of jobs However, the picture (across both public and private sectors) is now starting to look not only more nuanced but also more positive. A 2017 report from consultancy PWC suggested that embedding AI across all sectors is likely to create thousands of jobs. In the UK, one estimate suggests that it could contribute as much as 5% of GDP within 10 years. That’s not to say that we won’t lose jobs, because we undoubtedly will. However, they will be

Eyes are important, don't get me wrong. So are ears, noses, tongues, fingers, balance calibration organs and everything else that feeds that massive brain of yours.1 Salinity detectors in narwhals, electrical sensors in freshwater bottom feeders, echolocation in bats all provide sensory input that humans couldn't adequately process. Every beast has its own senses relevant to its own living conditions. Even your smartphone has cameras, microphones, gyroscopes, an accelerometer, a magnetometer, interfaces for phone/GPS/Bluetooth/WiFi, and some have a barometer, proximity sensors, and ambient light sensors. Biometric sensing equipment in today's phones can include optical, capacitive or ultrasonic fingerprint readers and an infrared map sensor for faces.

The European Union is considering banning the use of artificial intelligence for a number of purposes, including mass surveillance and social credit scores. This is according to a leaked proposal that is circulating online, first reported by Politico, ahead of an official announcement expected next week. If the draft proposal is adopted, it would see the EU take a strong stance on certain applications of AI, setting it apart from the US and China. Some use cases would be policed in a manner similar to the EU's regulation of digital privacy under GDPR legislation. Member states, for example, would be required to set up assessment boards to test and validate high-risk AI systems.

A European Union plan to regulate artificial intelligence could see companies that break proposed rules on mass surveillance and discrimination fined millions of euros. Draft legislation, leaked ahead of its official release later this month, suggests the EU is attempting to find a "third way" on AI regulation, between the free market US and authoritarian China. The draft rules represent an outright ban on AI designed to manipulate people "to their detriment", carry out indiscriminate surveillance or calculate "social scores". Much of the wording is currently vague enough that it could cover the entire advertising industry or nothing at all. In any case, the military and any agency ensuring public security are exempt.

This Position Statement presents a broad overview of issues arising at the intersection of AI and IP law based on the work of the Max Planck Institute for Innov

Several social factors impact how people respond to AI explanations used to justify AI decisions affecting them personally. In this position paper, we define a framework called the \textit{layers of explanation} (LEx), a lens through which we can assess the appropriateness of different types of explanations. The framework uses the notions of \textit{sensitivity} (emotional responsiveness) of features and the level of \textit{stakes} (decision's consequence) in a domain to determine whether different types of explanations are \textit{appropriate} in a given context. We demonstrate how to use the framework to assess the appropriateness of different types of explanations in different domains.