[{"id":541,"title":"AutoLab: Can an AI Replace Human Data Collectors?","sha":"0ef39e22b61491bdb64d4127d53668c4","user_id":null,"state":"published","body":"Can an autonomous AI conduct experiments on real humans in a lab safely and efficiently? With [Automated Laboratory (AutoLab)](https://devpost.com/software/autolab#updates), our open-source ubiquitous computing project, we will do just that. Imagine a Star Trek-like computer infused into a lab room. Participants enter the smart lab by activating an electronic lock, converse naturally with an AI during the study, and are compensated digitally. This will greatly speed up and standardize data collection in the health and social sciences. Currently, unpaid and overworked assistants must spend inordinate time and money to collect data. This results in human errors that waste tax dollars, lost time that could be spent on more creative tasks, and small datasets that limit generalizability to the larger population.\r\n\r\nTypical human experiments in the socio-behavioral and health sciences span 1-2 hours and need one more hour to prep and manage data. For sample sizes of 50-200, this requires at least 100-600 hours for one study (~ $2,000-12,000 tax dollars annually). Not long ago, statistical/visualization software, word processors, and communication tools advanced science by standardizing tasks, reducing errors, increasing speed, and fostering more creative pursuits. Ubiquitous computing can do the same by collecting large datasets from human subjects in social and health science labs 24/7/365.","created_at":"2018-12-08T00:07:27.531Z","updated_at":"2018-12-09T06:01:18.768Z","zenodo_id":2092676,"doi":"https://doi.org/10.5281/zenodo.2092676","subject":null,"tags":["ai","artificial intelligence","research cycle","internet of things","data collection","open source","productivity","software","ubiquitous computing","ubicomp"],"vote_count":0,"view_count":1,"score":0.1,"deleted":false,"muted":false,"tweeted":false,"attachment_file_name":null,"attachment_content_type":null,"attachment_file_size":null,"attachment_updated_at":null},{"id":89,"title":"On centralized vs. decentralized neurotechnologies","sha":"80fe103e7e0fe456a227a195745b2a8b","user_id":335,"state":"published","body":"Neurotechnology is any technology developed to interface with brain function. It is a rapidly advancing area with important implications for the study of the brain and neurological disorders, as well as consumerization of the connectome and related neurocognitive processes. However, the topology of different neurotechnologies remains limited. One useful classification is between centralized and decentralized neurotechnologies. Centralized neurotechnologies are defined as having a direct input and/or output with the central nervous system (CNS). An example of this is a deep brain stimulation machine with inputs to the cortex or subcortical nuclei that is used to modulate neurotransmitter activity. Conversely, decentralized neurotechnologies do not share a direct link to CNS, but are designed to sense and/or modulate neural function. An example is a machine learning algorithm that uses realtime biological signals to predict seizure onset in epileptics and allow for preventative actions. Here, it is hypothesized that centralized and decentralized neurotechnologies vary along two important axes: the power to modulate neural function and the invasiveness of the technology. This classification system has important implications for (a) estimating the rate of neurotechnological proliferation and (b) generating ethical guidelines for both scientific research programs and consumer product development.","created_at":"2015-02-22T20:27:22.392Z","updated_at":"2015-02-25T20:42:10.984Z","zenodo_id":23310,"doi":"http://dx.doi.org/10.5281/zenodo.15594","subject":null,"tags":["neurotechnology","neuroscience","ai","brain","technology"],"vote_count":1,"view_count":1,"score":1.1,"deleted":false,"muted":false,"tweeted":true,"attachment_file_name":null,"attachment_content_type":null,"attachment_file_size":null,"attachment_updated_at":null},{"id":797,"title":"Applying Machine Learning to Detect Code Quality Issues","sha":"b6bdcb6121c98e1d1a0bc0cc5870135d","user_id":null,"state":"published","body":"Detecting the potential problems in the code before the product is released can prevent the problems in production and lower the cost of the system operation. The automated code review tools are relying on detecting code patterns that are know to cause problems. This methods are unable to find new type of problems.\r\nWe can apply machine learning to the problem. The question would be how to find a good training set. The answer to this question would be: the public open source repositories (GitHub, Bitbucket, etc.). With each commit, the developer is providing the code fragment (the modifications of the original file) and the information about the change (fix, new feature, merge, etc.). \r\nWe can train model that will use the defect fixes to learn the pattens in which the code can be broken. We can extract the features from the code based on NLP, however we can go a step further and enhance the features. Since the code follows strict syntactical rules, we can generate features based on the parsed code tree (ex. level of nesting, recursion, etc.).\r\nThe model would return sections of the code that have high probability of having defect fixes in future. \r\n","created_at":"2021-01-28T15:48:08.006Z","updated_at":"2021-05-21T06:01:19.679Z","zenodo_id":4775046,"doi":"https://doi.org/10.5281/zenodo.4775046","subject":null,"tags":["ml","ai","machine learning","code quality"],"vote_count":0,"view_count":1,"score":0.1,"deleted":false,"muted":false,"tweeted":false,"attachment_file_name":null,"attachment_content_type":null,"attachment_file_size":null,"attachment_updated_at":null},{"id":1173,"title":"Alternative to Turing’s test on AI","sha":"23e327847db8f0e50124da45ec0aa751","user_id":null,"state":"published","body":"As Turing wrote in his Can digital computers think?, we could know if a subject not of our own kind is an intelligent source through their output signals. However, the signal he selected is perhaps not as clear cut as he presupposed: “… it will be possible to programme a machine to answer questions in such a way that it will be extremely difficult to guess whether the answers are being given by a man or by the machine.” This future is here with ever-advancing AI tools already able to build convincing answers. Von Neuman suggested a stricter test: “the sudden pounce on an idea.\r\nI suggest intonation—reading aloud—as a test of an intelligent subject. Ask any text-to-speech software  to recite Lorca’s Lament for Ignacio Sanchez Mejias or Elliot’s The Waste Land. Intonation is contingent on meaning, which is a manifestation of understanding.\r\nA real test to detect a human signal producing-being could be the congruent emotional reading of a new poem or a paragraph for which meaning, context and understanding is essential. It is not a test of emotional response á la Blade Runner’s Voight-Kampff test, but one that relies on evaluating emotional intonation output for a diagnosis..\r\n","created_at":"2023-11-29T07:42:51.276Z","updated_at":"2024-04-14T16:11:30.337Z","zenodo_id":10426525,"doi":"https://doi.org/10.5281/zenodo.10426525","subject":null,"tags":["ai","meaning","context","understanding","intonation"],"vote_count":3,"view_count":1,"score":3.1,"deleted":false,"muted":false,"tweeted":false,"attachment_file_name":null,"attachment_content_type":null,"attachment_file_size":null,"attachment_updated_at":null},{"id":1399,"title":"Capitalist capital allocation as meta-reward function for AI training model","sha":"cecf7e1646a5f412b2c60014d226ef87","user_id":null,"state":"published","body":"Machine learning protocols utilize rewards function during training as a means of tuning parameters toward obtaining desirable outputs from a model. One challenge for the current AI industry is the difficulty of translating real-world utility into reward functions for individual models that are trained primarily on training sets of online content, which seldom has an accurate valuation of its real-world utility attached to it. However, computing resources are allocated to different models based on their value to the companies producing and, at a higher layer, capital in the form of investment or revenue is preferentially allocated to those companies whose models are perceived as having higher potential for real-world utility. In essence, this process functions as a meta-reward function for the AI ecosystem overall.  ","created_at":"2024-07-24T10:18:55.003Z","updated_at":"2024-07-30T06:00:36.428Z","zenodo_id":13129560,"doi":"https://doi.org/10.5281/zenodo.13129560","subject":null,"tags":["ai","machine learning","economics"],"vote_count":0,"view_count":1,"score":0.1,"deleted":false,"muted":false,"tweeted":false,"attachment_file_name":null,"attachment_content_type":null,"attachment_file_size":null,"attachment_updated_at":null}]