Cloud Hosting

An artificial intelligence program has developed its own language and no one can understand it.

OpenAI is anartificial intelligencesystems developer their programs are fantastic examples of super-computing but there are quirks.

DALLE-E2 isOpenAIs latest AI system it can generate realistic or artistic images from user-entered text descriptions.

DALLE-E2 represents a milestone in machine learning OpenAIs site says the program learned the relationship between images and the text used to describe them.

A DALLE-E2 demonstration includesinteractive keywordsfor visiting users to play with and generate images toggling different keywords will result in different images, styles, and subjects.

But the system has one strange behavior itswritingits own language of random arrangements of letters, and researchers dont know why.

Giannis Daras, a computer science Ph.D. student at the University of Texas, published aTwitter threaddetailing DALLE-E2s unexplained new language.

Daras told DALLE-E2 to create an image of farmers talking about vegetables and the program did so, but the farmers speech read vicootes some unknown AI word.

Darasfedvicootes back into the DALLE-E2 system and got back pictures of vegetables.

We then feed the words: Apoploe vesrreaitars and we get birds. Daras wrote on Twitter.

It seems that the farmers are talking about birds, messing with their vegetables!

Stay up on the very latest with Evening Update.

Daras and a co-author have written apaperon DALLE-E2s hidden vocabulary.

They acknowledge that telling DALLE-E2 to generate images of words the command an image of the word airplane is Daras example normally results in DALLE-E2 spitting out gibberish text.

When plugged back into DALLE-E2, that gibberish text will result in images of airplanes which says something about the way DALLE-E2 talks to and thinks of itself.

Some AI researchers argued that DALLE-E2s gibberish text is random noise.

Hopefully, we dont come to find the DALLE-E2s second language was a security flaw that needed patching after its too late.

This article originally appeared onThe Sunand was reproduced here with permission.

See the original post:
Artificial intelligence spotted inventing its own creepy language - New York Post

The following article is an opinion piece written by Mark J. Kiel. The views and opinions expressed in this article are those of the author and do not necessarily reflect the official position of Technology Networks.

The recent publication of the gapless, telomere-to-telomere human genome assembly serves as a real reminder of the significant work that remained to be completed from initial draft genome to the final sequence assembly. Similarly, while we have learned much about causative genetic variants in human disease over the past many years, there is much work that remains to complete this knowledge and to put it into practice.

To fully realize the promise of precision medicine, it is my contention that we must pre-curate every variant in the human genome. It is not scalable for analysts in clinical practice to spend as much as 90 minutes per variant to report the results of a molecular lab test. Pre-curation will make this analysis maximally efficient and reproducible.

A great example of where pre-curation will be beneficial is newborn screening by next-generation sequencing. This new method of screening demands a great deal of information about a multitude of diseases to ensure the utmost accuracy of resulting diagnoses. However, real-time assessment of this information for each patient and each variant will challenge the scalability of this initiative.

Conventional methods for variant curation are laborious, slow, incomplete and error-prone relying as they do on painstaking manual searches for evidence in the scientific and clinical literature. These manual processes too often miss key data and lead to inaccurate conclusions about the clinical significance of a patients variant. Collectively, all of this previous work has led to the pre-curation of just a fraction of human genetic variants. Moreover, these older pre-curations are often inadequate as new knowledge is being created every day requiring continuous updates to existing curations to ensure nothing is missed. If we continue to rely on outmoded techniques, it will not be possible to fully curate the human genome in our lifetimes.

Faster and more comprehensive variant curation will require a combinatorial approach merging the scalability and sensitivity of AI and the specificity and accuracy that can only come from the expert judgment of experienced human curators. AI-driven indexing of the scientific and clinical literature can ensure more complete information for each variant, while AI-informed expert curation delivers maximum specificity for results in a maximally efficient manner.

The primary bottleneck to achieving accurate and comprehensive variant curation is the need to manually locate, assess, annotate and document evidence from the scientific and clinical literature. In particular, the nuances of the genetic code and the idiosyncrasies of genetic nomenclature and other complexities of biology make it difficult to disambiguate terminology and ensure that curations are correct and complete.

AI is a natural solution to these challenges. Only computational approaches can meet the scale and sensitivity requirements of this ambitious genome-wide variant curation. There is precedent for using AI to index vast amounts of data often unstructured and poorly organized data so it is an excellent fit for indexing the scientific and clinical literature. Paying attention to and resolving genetic ambiguities and focusing on the most critical clinical information is paramount.

A team of highly trained experts is necessary to carefully assess the assembled evidence and make an informed judgment about its appropriateness and applicability, as well as ensure the utmost accuracy of all final interpretations. This review process can be further accelerated by AI-driven organization and annotation of the data.

This approach is already making it possible to deliver new insights about disease-causing variants in patients. In one example, scientists at the Rare Genomics Institute reanalyzed previously inconclusive exome results using an AI-powered tool and found a single scientific report that allowed them to classify a key variant as pathogenic and produce an evidence-based diagnosis and effective treatment for the patient.

With the speed and scale afforded by AI technology, it will be possible within the next several years to curate the entire human genome. With a combinatorial approach that brings together the best of automated AI and expert curation, we can firmly establish the foundation of genomic intelligence needed to make precision medicine a reality for all patients.

About the author:

Mark Kiel is chief scientific officer and co-founder of Genomenon, an AI genomics company. He has extensive experience in genome sequencing and clinical data analysis. Mark is a molecular genetic pathologist having received his MD/PhD in stem cell biology and cancer genomics from the University of Michigan.

Continue reading here:
Curation of the Entire Human Genome Requires the Best of Both Human and Artificial Intelligence - Technology Networks

Greece is just one example of a population where the share of older people is expanding, and with it the incidences of neurodegenerative diseases. Among these, Alzheimers disease is the most prevalent, accounting for 70% of neurodegenerative disease cases in Greece. According to estimates published by the Alzheimer Society of Greece, 197,000 people are suffering from the disease at present. This number is expected to rise to 354,000 by 2050.

Dr. Andreas Papadopoulos1, a physician and scientific coordinator at Iatropolis Medical Group, a leading diagnostic provider near Athens, Greece, explains the key role of early diagnosis: The likelihood of developing Alzheimers may be only 1% to 2% at age 65. But then it doubles every five years. Existing drugs cannot reverse the course of the degeneration; they can only slow it down. This is why its crucial to make the right diagnosis in the preliminary stageswhen the first mild cognitive disorder appearsand to filter out Alzheimers patients2.

Diseases like Alzheimers or other neurodegenerative pathologies characteristically have a very slow progression, which makes is difficult to recognize and quantify pathological changes on brain MRI images at an early stage. In evaluating scans, some radiologists describe the process as one of guesstimation, as visual changes in the highly complex anatomy of the brain are not always possible to observe well with the human eye. This is where technical innovations such as artificial intelligence can offer support in interpreting clinical images.

One such tool is the AI-Rad Companion Brain MR3. Part of a family of AI-based, decision-support solutions for imaging, AI-Rad Companion Brain MR is a brain volumetry software that provides automatic volumetric quantification of different brain segments. It is able to segment them from each other: it isolates the hippocampi and the lobes of the brain and quantifies white matter and gray matter volumes for each segment individually. says Dr. Papadopoulos. In total, it has the capacity to segment, measure volumes, and highlight more than 40 regions of the brain.

Calculating volumetric properties manually can be an extremely laborious and time-consuming task. More importantly, it also involves a degree of precise observation that humans are simply not able to achieve. says Dr. Papadopoulos. Papadopoulos has always been an early adopter and welcomed technological innovations in imaging throughout his career. This AI-powered tool means that he can now also compare the quantifications with normative data from a healthy population. And its not all about the automation: the software displays the data in a structured report and generates a highlighted deviation map based on user settings. This allows the user to also monitor volumetric changes manually with all the key data prepared automatically in advance.

Opportunities for more accurate observation and evaluation of volumetric changes in the brain encourages Papadopoulos when he considers how important the early detection of neurodegenerative diseases is. He explains: In the early stages, the volumetric changes are small. In the hippocampus, for example, there is a volume reduction of 10% to 15%, which is very difficult for the eye to detect. But the objective calculations provided by the system could prove a big help.

The aim of AI is to relieve physicians of a considerable burden and, ultimately, to save time when optimally embedded in the workflow. An extremely valuable role for this particular AI-powered postprocessing tool is that it can visualize a deviation of the different structures that might be hard to identify with the naked eye. Papadopoulos already recognizes that the greatest advantage in his work is the objective framework that AI-Rad Companion Brain MR provides on which he can base his subjective assessment during an examination.

AI-Rad Companion4 from Siemens Healthineers supports clinicians in their daily routine of diagnostic decision-making. To maintain a continuous value stream, our AI-powered tools include regular software updates and upgrades that are deployed to the customers via the cloud. Customers can decide whether they want to integrate a fully cloud-based approach into their working environment leveraging all the benefits of the cloud or a hybrid approach that allows them to process imaging data within their own hospital IT setup.

The upcoming software version of AI-Rad Companion Brain MR will contain new algorithms that are capable of segmenting, quantifying, and visualizing white matter hyperintensities (WMH). Along with the McDonald criteria, reporting WHM aids in multiple sclerosis (MS) evaluation.

Read the original:
Evaluating brain MRI scans with the help of artificial intelligence - MIT Technology Review

New Delhi: Youve probably heard for years that the workforce would be supplanted by robots. AI has changed several roles, such as using self-checkouts, ATMs, and customer support chatbots. The goal is not to scare people, but to highlight the fact that AI is constantly altering lives and executing activities to replace the human workforce. At the same time, technological advancements are producing new career prospects. AI is predicted to increase the demand for professionals, particularly in robotics and software engineering. As a result, AI has the potential to eliminate millions of current occupations while also creating millions of new ones.

Among the many concerns that AI raises is the possibility of wiping out a large portion of the human workforce by eliminating the need for manual labour. But it will simultaneously liberate humans from having to perform tedious, repetitive tasks, allowing them to focus on more complex and rewarding projects, or simply take some much-needed time off.

According to a McKinsey report, depending on the adoption scenario, automation will displace between 400 and 800 million jobs by 2030, requiring up to 375 million people to change job categories entirely.

These figures can make people feel uneasy and anxious about the future. However, history suggests that this may not be the case; there is no question that some industries will be transformed to the point where they no longer require human labour, leading toward job redefinition and business process reform. For eg, the diagnosis of many health issues could be effectively automated, making doctors focus on other major issues that need their attention. In terms of replacing humans completely, human labour is and will continue to be necessary for the foreseeable future.

Though the potential of AI is unimaginable it is also restricted. While it is apparent that AI will dominate the professional world on many levels. However, there can be no denial that as advanced as AI may be, it can and never will be able to replicate human consciousness that reinforces the human beings position at the top of the food chain.

Until now, we are talking about the jobs that can be snatched as technology advances but then, the human aspects of work cannot be replaced. Lets focus on something that they cannot do. There are some jobs that only humans are capable of performing.

There are jobs that require creation, conceptualization, complex strategic planning, and dealing with unknown spaces and feelings or emotional interactions that are way beyond the expertise of an AI as of now. Lets now talk about certain skills that are irreplaceable till the human race exist.

1. Empathy is unique to humans: Some may argue that animals show empathy as well, but they are not the ones taking over the jobs. Humans, unlike programmed software designed to produce a specific result, are capable of feeling emotions. It may seem contradictory, but the personal affinity between a person and an organisation is the foundation of a professional relationship. Humans need a personal connection that extends beyond the professional realm to develop trust and human connection, something that bot technology completely lacks.

2. Emotional Intelligence: Though accurate, the AI is not intuitive, or culturally sensitive because thats a human trait. No matter how accurately it is programmed to carry out a task, it cannot possess the human ability to adjust to the algorithm of human intellect. For instance, reading into the situation or the face of another human. It lacks emotional intellect which makes humans capable of understanding and handling an interaction that needs emotional communication. Exactly during your customer care service, one would always prefer a human interaction to read and understand the situation than an automated machine that cannot work or help beyond the programming.

3. Creativity: Perk of being human: AI can improve productivity and efficiency by reducing errors and repetition and replacing manual jobs with intelligent automated solutions, but it cannot comprehend human psychology. Furthermore, as the world becomes more AI-enabled, humans will be able to take on increasingly innovative tasks.

4. Problem-solving outside a code: Humans can deal with unexpected uncertainty by analysing the situation; like critical thinking during complex scenarios, and adopting creative tactics. Unlike humans, who may function under a variety of obstacles and settings, AI-powered devices cannot perform beyond their function, may be far in the future but not in the foreseeable one.

There is not even the slightest doubt that AI will not drive the future. To make AI work, humans need to be creative, insightful, and contextually aware. The reason for this is straightforward: humans will continue to provide value that machines cannot duplicate.

Continued here:
4 skills that won't be replaced by Artificial Intelligence (AI) in the future - Kalinga TV

A new study finds that an artificial intelligence model can predict whether Crohns disease will recur after surgery.

A deep learning model trained to analyze histological images of surgical specimens accurately classified patients with and without Crohns disease recurrence, investigators report in The American Journal of Pathology.

According to researchers, more than 500,000 individuals in the United States have Crohns disease. Crohns disease is a chronic inflammatory bowel disease that damages the digestive system lining. It can cause digestive system inflammation, which may result in abdominal pain, severe diarrhea, exhaustion, weight loss, and malnutrition.

Many people end up needing surgery to treat their Crohns disease. Even after a successful operation, recurrence is common. Now, researchers are reporting that their AI tool is highly accurate at predicting the postoperative recurrence of Crohns disease. It also linked recurrence with the histology of subserosal adipose cells and mast cell infiltration.

Using an artificial intelligence (AI) tool that simulates how humans visualize and is trained to identify and categorize pictures, researchers created a model that predicts the postoperative recurrence of Crohns disease with high accuracy by evaluating histological images. The AI tool also identified previously unknown differences in adipose cells and substantial disparities in the degree of mast cell infiltration in the subserosa, or outer lining of the gut, when comparing individuals with and without disease recurrence. Elseviers The American Journal of Pathology published the findings.

The 10-year rate of postoperative symptomatic recurrence of Crohns disease, a chronic inflammatory gastrointestinal illness, is believed to be 40%. Although there are scoring methods to measure Crohns disease activity and the existence of postoperative recurrence, no scoring system has been devised to predict whether Crohns disease will return.

Sixty-eight patients with Crohns disease were classified according to the presence or absence of postoperative recurrence within two years. The investigators performed histological analysis of surgical specimens using deep learning EfficientNet-b5, a commercially available AI model designed to perform image classification. They achieved a highly accurate prediction of postoperative recurrence (AUC=0.995) and discovered morphological differences in adipose cells between the two groups. Credit: The American Journal of Pathology

Most of the analysis of histopathological images using AI in the past have targeted malignant tumors, explained lead investigators Takahiro Matsui, MD, Ph.D., and Eiichi Morii, MD, Ph.D., Department of Pathology, Osaka University Graduate School of Medicine, Osaka, Japan. We aimed to obtain clinically useful information for a wider variety of diseases by analyzing histopathology images using AI. We focused on Crohns disease, in which postoperative recurrence is a clinical problem.

The research involved 68 Crohns disease patients who underwent bowel resection between January 2007 and July 2018. They were divided into two groups based on whether or not they had postoperative disease recurrence within two years after surgery. Each group was divided into two subgroups, one for training and the other for validation of an AI model. Whole slide pictures of surgical specimens were cropped into tile images for training, labeled for the presence or absence of postsurgical recurrence, and then processed using EfficientNet-b5, a commercially available AI model built to perform image classification. When the model was tested with unlabeled photographs, the findings indicated that the deep learning model accurately classified the unlabeled images according to the presence or absence of disease occurrence.

Following that, prediction heat maps were created to identify areas and histological features from which the machine learning algorithm could accurately predict recurrence. All layers of the intestinal wall were shown in the photos. The heatmaps revealed that the machine learning algorithm correctly predicted the subserosal adipose tissue layer. However, the model was less precise in other regions, such as the mucosal and proper muscular layers. Images with the greatest accurate predictions were taken from the non-recurrence and recurrence test datasets. The photos with the greatest predictive results all had adipose tissue.

Because the machine learning model achieved accurate predictions from images of subserosal tissue, the investigators hypothesized that subserosal adipose cell morphologies differed between the recurrence and the non-recurrence groups. Adipose cells in the recurrence group had a significantly smaller cell size, higher flattening, and smaller center-to-center cell distance values than those in the nonrecurrence group.

These features, defined as adipocyte shrinkage, are important histological characteristics associated with Crohns disease recurrence, said Dr. Matsui and Dr. Morii.

The investigators also hypothesized that the differences in adipocyte morphology between the two groups were associated with some degree or type of inflammatory condition in the tissue. They found that the recurrence group had a significantly higher number of mast cells infiltrating the subserosal adipose tissue, indicating that the cells are associated with the recurrence of Crohns disease and the adipocyte shrinkage phenomenon.

To the investigators knowledge, these findings are the first to link postoperative recurrence of Crohns disease with the histology of subserosal adipose cells and mast cell infiltration. Dr. Matsui and Dr. Morii observed, Our findings enable stratification by the prognosis of postoperative Crohns disease patients. Many drugs, including biologicals, are used to prevent Crohns disease recurrence, and proper stratification can enable more intensive and successful treatment of high-risk patients.

Reference: Deep Learning Analysis of Histologic Images from Intestinal Specimen Reveals Adipocyte Shrinkage and Mast Cell Infiltration to Predict Postoperative Crohn Disease by Hiroki Kiyokawa, Masatoshi Abe, Takahiro Matsui, Masako Kurashige, Kenji Ohshima, Shinichiro Tahara, Satoshi Nojima, Takayuki Ogino, Yuki Sekido, Tsunekazu Mizushima and Eiichi Morii, 28 March 2022, The American Journal of Pathology.DOI: 10.1016/j.ajpath.2022.03.006

See original here:
Artificial Intelligence Model Can Successfully Predict the Reoccurrence of Crohns Disease - SciTechDaily