"Artificial intelligence learned how to generate text and art better than ever before, while computer scientists developed algorithms that solved long-standing problems." Links to further papers and discussion on topics including: Tackling "P versus NP" Emergent behaviours in large language models...
Source: quantamagazine.org
We've reached the Skynet moment for AI, it would seem. GPT-4, cast in the role of Terminator, has not only achieved consciousness, but it's coming for
Source: datanami.com
A new generation of artificial intelligence-based information access systems, which includes Microsoft’s Bing/ChatGPT, Google/Bard and Meta/LLaMA, is upending the traditional search engine mode of search input and output. These systems are able to take full sentences and even paragraphs as input and...
Source: ampproject.org
This pre-print paper suggests ChatGPT could change how assessments might be done. “We evaluated the performance of a large language model called ChatGPT on the United States Medical Licensing Exam (USMLE), which consists of three exams: Step 1, Step 2CK, and Step 3. ChatGPT performed at or near the...
A group of over 1,000 AI researchers has created a multilingual large language model bigger than GPT-3—and they’re giving it out for free.
Source: technologyreview.com
The company believes its Codex machine learning algorithm is the next step in programming—a sidekick for coders to speed up the work and ease the drudgery.
Source: singularityhub.com
Despite an increased focus and urgency for CE/CME professionals to effectively and systematically assess the impact of their educational interventions, the community has struggled to do so. This struggle is in large part due to the lack of a standardised outcomes language and a set of unified approaches.
Source: nih.gov
Last summer I blogged about using a Deep Neural Network to generate tweets but only used 3200 of my tweets. Since then I've used Twitter's archive mechanism to retrieve ALL my tweets (just over 30,000) to train a network. Not any old network - the GPT-2 model from OpenAI. This 'finetuning' of an existing...
An article from MIT Technology Review showing how the World Food
Programme uses geospatial data that is developed and made 'open' to all
by people within the areas being served. "It’s one of the most beautiful places on Earth, but its people are among the most vulnerable. Afghanistan’s snowy...
Source: technologyreview.com
"Steph Norman and Aaron Willoughby cannot have their ceremony entirely in Cornish." Apparently they could conduct it English obviously but also in Welsh (which is close to Cornish) but it highlights the lack of legal formality to the Cornish language.
Source: bbc.com
At age 30, the Python programming language has never been used by more developers across more use cases than it is today.
Source: venturebeat.com
Scaling Kubernetes to 7,500 Nodes: We've scaled Kubernetes clusters to 7,500 nodes, producing a scalable infrastructure for large models like GPT-3, CLIP, and DALL·E, but also for rapid small-scale iterative research such as Scaling Laws for Neural Language Models. Scaling a single Kubernetes cluster...
Source: openai.com
Ever wondered what an armchair in the shape of an avocado might look like? Introducing Open-AI's DALL-E.
Does this help with accessibility by explaining things in pictures from written words? Does it risk replacing humans in the creative industry with machines?
"DALL·E: Creating Images from...
Source: openai.com
Supporting migrants and refugees with posttraumatic stress disorder: development, pilot implementation, and pilot evaluation of a continuing interprofessional education for healthcare providers: Refugees and migrants face an increased risk of developing posttraumatic stress disorder (PTSD). Adequate...
Source: biomedcentral.com
Learning to Summarize with Human Feedback: We've applied reinforcement learning from human feedback to train language models that are better at summarization. Our models generate summaries that are better than summaries from 10x larger models trained only with supervised learning. Even though we train...
Source: openai.com
Comparison of E-Learning, M-Learning and Game-based Learning in Programming Education – A Gendered Analysis: Learning to code is considered as a difficult and challenging task for a significant number of novice programmers in programming education. Novice programmers have to acquire different skills...
Source: online-journals.org
My next tweet could be generated by a deep-learning neural network. I've been training one. Would anyone notice the difference? Could I just hand over tweeting to my machine? Method: downloaded the last 3200 Tweets that I posted using allmytweets.netpruned the dates off and removed the RTs by using some...
Source: agnate.co.uk
How To Kill Ideas: We were asked last week by the Disruptive Innovators Network, 'How long should you spend on an idea?' "In the early days of Bromford Lab we had a 12 WEEKS MAX rule. If we couldn’t get an idea up and running within that time – it should be killed. We soon realised the error of our...
Source: paulitaylor.com
Could deep learning AI be used to screen TIA clinic referral letters? Possibly. Stroke prevention clinics (TIA or Transient Ischaemic Attack) clinics are an important aspect of urgent care. A TIA is a risk factor for future stroke. Medical treatment needs to be started quickly and surgical options, if...
Source: agnate.co.uk
Communication Strategies for Sharing Prognostic Information With Patients—Beyond Survival Statistics: This Viewpoint discusses the challenges of communicating uncertain prognosis to patients and offers a framework and language that go past mortality statistics—addressing range of time left, loss...
Source: jamanetwork.com