Waste management is essential as the world combats climate change and environmental destruction. Artificial intelligence is supporting waste management through training ML models that detect materials including plastics, glass, cans, and newspapers.
Estimates show that approximately 2 billion tons of waste produced create negative environmental implications. Greyparrot is supporting these efforts by use of computer vision to facilitate waste materials detection and recycling.
Microsoft is transitioning to AI-based curation of articles for its news website and replacing humans. Despite the grim news here, Microsoft will slash only 50 jobs but only those repetitive in content curation. The AI applications used by Microsoft for their news produce better results and offer cool features such as photo recommendations for their editors.
The human factor remains vital in AI deployment and the C-Suite should ask questions such as the organization of technology and value across the enterprise. Humanization of technology determines the success of AI deployment in organizations and business executives should consider these options.
These and more insights on our Weekly AI Update
Microsoft using AI to curate Articles
Microsoft is deciding to lay off dozens of journalists and editorial workers. The layoffs reveal Microsoft’s bigger push to rely on artificial intelligence to pick news and content that’s presented on MSN, inside Microsoft’s Edge browser, and in the company’s various Microsoft News apps.
50 jobs are affected in the US due to the layoff. The job losses are also affecting international teams, and The Guardian reports that around 27 are being let go in the UK after Microsoft’s decision to stop employing humans to curate articles.
After launching MSN all the way back in 1995, Microsoft has been in the news business for more than 25 years. It had “more than 800 editors working from 50 locations around the world for its MicrosoftNews.
Microsoft has been moving towards AI gradually for its Microsoft News work to encourage publishers and journalists to make use of AI. Microsoft has been using AI to scan for content process¹ and filter and even suggest photos for human editors to pair it with.
Satelite Imagery for Environmental Management
Over the last few years as global temperatures rise and weather patterns change the U.S. has suffered from devastating wildfires, making the otherwise natural phenomenon especially unpredictable and severe.
Stanford researchers have found a way to track and predict risky areas to help out using #machinelearning and satellite imagery².
The way forests and shrublands are currently being tested for susceptibility to wildfires is by manually collecting branches and foliage. Quite labor-intensive and difficult activities required to scale it accurate and reliable.
Sentinel and Landsat satellites have amassed a trove of imagery of the Earth’s surface that, when carefully analyzed, could provide a secondary source for assessing wildfire risk — and no one has to risk getting splinters for.
The Sentinel satellites’ “synthetic aperture radar,” can pierce the forest canopy and image the surface below leveraged by the Stanford team.
Stanford hydrologist Alexandra Konings said “One of our big breakthroughs was to look at a newer set of satellites that are using much longer wavelengths, which allows the observations to be sensitive to water much deeper into the forest canopy and be directly representative of the fuel moisture content,”
Demand for Data
The companies that enable artificial intelligence have multiplied and in many cases prospered as AI³ has grown from niche to mission-critical technology.
Just in a couple of years, DefinedCrowd has transformed from the Disrupt stage to globe-spanning AI toolkit to the Fortune500. To further fuel its expansion the company just raised a new $50.5 million B round.
DefinedCrowd¹¹ specialized in #naturallanguageprocessing supplies data used to create AI. It would be much more difficult for machine learning systems to tell what users mean if someone had to vet the 500 different ways you could ask for the weather.
There’s been no shortage of companies in search of training data as AI has worked its way into everything from creating and editing media to enterprise software.
CEO and Co-founder Daniela Braga told TechCrunch.“The demand for data has consistently been growing over the last couple years — companies are more and more aware of the impact that data has on their systems, and have been looking for more languages and domains that weren’t considered five years ago,”
New markets and applications are opening up constantly and need high-quality data to develop consumer-ready products. the company reported a tremendous 656% increase in revenue year-over-year as evidence.
The Human Factor in AI Adoption
Many companies are accelerating the push for AI tools that can keep their employees safer and more connected — while also keeping their customers happier and more satisfied as the coronavirus continues to wreak havoc on businesses around the world. It’s essential for business leaders to take the time to consider the human factor in AI adoption⁴.
Humans must be willing to use the technology if it’s going to be successful. There are two potential “problem areas” in any AI rollout even without the coronavirus emergency. How AI was organized? Was it created to scale across your enterprise?
The businesses rising amidst this chaotic time are using AI Technology powered by chips, frameworks, infrastructure, and software from companies like NVIDIA, AWS, Intel, Oracle, IBM, etc.
AI can also help make nearly any job more remote and reduce the health risks of person-to-person interaction. A surge in AI deployment is in demand clearly for pandemic preparedness.
AI and ML Pushing Technology Boundaries
Artificial intelligence, machine learning, NLP, and #computervision are quite familiar to those who put work into technology.
AI/ML will play an integral role in shaping this tech era’s most successful business models as the boundaries of these technologies are constantly being pushed and broadened.
The machine learning market is poised to more than quadruple in the coming years.
When used to good effect AI/ML solutions can equip your organization with a significant competitive advantage. Every technology vendors now offer AI/ML services but if anything we are often inundated with choices in this domain. How do we know we’re making the right choices to use these services to good effect?
It is simple, we need a lot of high-quality data as poor data is the biggest impediment to successfully adopting and deploying AI/ML solutions⁵. Take IBM’s Watson for oncology as a cautionary tale.
The Watson supercomputer was discovered to generate “erroneous cancer treatment advice” which ranged from incorrect to outright unsafe after being trained on a small number of synthetic cancer cases.
The data management process intrinsically linked to AI initiatives covers everything from data creation or acquisition to transmission and storage.
Game Design and Development using AI
Researchers at Electronic Arts are testing recent advances in artificial intelligence as a way to speed the development process and make games more lifelike.
The researchers are harnessing an AI technique that proves itself by playing some of the earliest console video games⁶.
The reinforcement learning technology which is loosely inspired by the way animals learn in response to positive and negative feedback to automatically animate humanoid characters being used by a team from EA and the University of British Columbia in Vancouver. Fabio Zinno, a senior software engineer at ElectronicArts says, “The results are very, very promising,”
Characters and actions in video games are traditionally crafted manually. Sports games, such as FIFA, make use of motion capture, a technique that involves tracking a real person often using markers on their face or body.
But the possibilities are limited by the actions that have been recorded, and code still needs to be written to animate the character.
AI could save game companies millions of dollars while making games more realistic and efficient by automating the animation process, as well as other elements of game design and development.
Leveraging Personal Data to combat COVID-19
South Korea’s innovative use of technology is credited as a critical factor in combating the spread of coronavirus. Many governments are turning to AI tools⁷ to both advance the medical research and manage public health whereas Europe and the United States struggle to cope.
AI tool is being leveraged for technical solutions such as contact tracing, symptom tracking, immunity certificates, and other applications.
Technologies are promising, but they must not undermine human rights.
Analyzing millions of data points from credit card transactions, CCTV footage, and cell phone geolocation data, Seoul has collected extensively and intrusively the personal data of its citizens to take action against the spreading of the pandemic.
The country’s Ministry of the Interior and Safety even developed a smartphone app that shares with officials GPS data of self-quarantined individuals. The app alerts officials when someone in quarantine crosses the “electronic fence” of their assigned area.
Many countries leveraged personal data combating the coronavirus. This data supercharged with #artificialintelligence and machine learning cannot only be used for social control and monitoring but also to predict travel patterns, pinpoint future outbreak hot spots, and model chains of infection.
AI powered by 5G for new Economic Growth Models
Data in 2020 and 2021 alone shows that, developing countries’ repayments on their public external debt alone will soar to between $2.6 trillion and $3.4 trillion according to United Nations.
The coronavirus is pushing 40 to 60 million people into extreme poverty according to the prediction of The World Bank.
AI powered by 5G is increasingly at the forefront as countries need new economic growth models. Companies and individuals need to seriously think and prepare about post Covid-19 world. New business models are harnessing more towards AI adoption because of coronavirus recession.
Over 80% of the daily moves in the US stock market and the most valuable companies in the world all use computational social science are now believed to be AI machine learning algorithm trading.
Several niche areas have already been affected dramatically by the initial development of AI, but it now seems to impact just about all parts of the economy, and indeed all aspects of our lives.
AI chatbots are zooming in to replace human call center workers⁸. As an advanced technology, AI is affecting both developed economies and developing economies. The developing economies are to focus on AI to strengthen the country’s industries.
Anew approach to neural architecture search⁹ (NAS), a technique that involves evaluating hundreds or thousands of AI models to identify the top performers has been proposed by researchers affiliated with Uber AI and OpenAI.
The technique called Synthetic PetriDish accelerates the most computationally intensive NAS steps while predicting model performance with higher accuracy than previous methods.
By testing candidate models’ overall performance, dispensing with manual fine-tuning NAS teases out top model architectures for tasks.
But a large amount of computation and data is required, the implication being that the best architectures train near the bounds of available resources.
To address this dilemma Synthetic Petri Dish takes an idea from biology. To create small models and evaluate them with generated data samples Synthetic Petri Dish uses candidate architectures.
Synthetic Petri Dish needs only a few performance evaluations of architectures and enables “extremely rapid” testing of new architectures.
While generating a set of architectures through an off-the-shelf NAS method the initial evaluations are used to train a Petri dish model. Synthetic Petri Dish required only a tenth of the original NAS’ compute.
AI Applications in Waste Management Industry
Greyparrot¹²- A London-based company uses computer vision AI to scale efficient processing of recycling has bagged £1.825 million (~$2.2M) in seed funding, topping up the $1.2M in pre-seed funding it had raised previously.
The early-stage European industrial techinvestor Speedinvest led the latest round with the participation from UK-based early stage b2b investor, Force Over Mass.
The Tech Crunch Disrupt SF battlefield alum has trained a series of machine learning models to recognize different types of waste, such as glass, paper, cardboard, newspapers, cans, and different types of plastics, in order to make sorting recycling more efficient, applying digitization and automation¹⁰ to the waste management industry.
Each year some 60% of the 2BN tonnes of solid waste produced globally ends up in open dumps and landfill, causing major environmental impact. While global recycling rates are just 14% — a consequence of inefficient recycling systems, rising labor costs, and strict quality requirements imposed on recycled material.
Less than 1% of waste is monitored and audited, per the startup currently. This AI applications that’re not so much taking over a human job as doing something humans essentially don’t bother with, to the detriment of the environment and its resources.
AI, Automation and Robotics in Operations
The mining technology and equipment company Komatsu’s Antoine Desmet shared how they’re using advanced forms of AI, automation, and robotics to make an impact on the organization’s operations on a recent AI Today podcast episode.
In the last few years, Antoine Desmet decided to start implementing machine learning and his company has seen significant growth through this approach, with his data science team growing from just one person to ten people.
The mining industry uses a lot of big, expensive machinery and much of this machinery has many sensors that provide large volumes of data that give insights into how the very expensive machines are operating, the conditions in which they operate, and also insights into their performance on specific tasks.
Advanced analytics prevents the need for humans to have to travel deep into the mine in potentially dangerous situations to try to evaluate problems and determine what is happening.
Moreover, #predictiveanalytics enables more strategic and efficient operations from maintenance to purchasing equipment. AI and machine learning can be applied to using drone footage or satellite imagery to keep a constant watch on waste and output piles to make sure that they are keeping in compliance with environmental regulations.
Solving Image Recognition Problems
Image recognition doesn’t only bring portrait mode on phones but it is also essential to self-driving cars and to identify tumors in healthcare. The problem with image recognition is typically solved via a machine learning algorithm.
They need to be trained on data to work properly
The data used to train many of the algorithms might not be sufficient, biased, or wrong. Often you won’t know If the data is wrong then the output might be wrong, and there is little control and oversight to detect those issues.
Machine learning algorithms’ insights are built on historic data. The algorithm can only replicate insights on the images. China used #facerecognition to spot who is jaywalking at a red spotlight using existing photos to train the algorithm to match faces.
Biased data is systematically missing data. If a situation has a low prevalence then the likelihood for this situation to occur is as well low.
ImageNet¹³ tries to tackle issues and asks people to label real images which are a common way to create a data set. Take images of people and ask thousands of people to label them.
Tools like AmazonMechanicalTurk or Upwork make labeling a very affordable task. People from all over the world label images for pennies a photo. But what are those labels?
¹AI to Scan for Content Process, ²Satelite Imagery, ³Artificial Intelligence, ⁴Human Factor in AI Adoption, ⁵Deploying AI/ML Solutions, ⁶Console Video Games, ⁷AI Tools, ⁸Human Call Center Workers, ⁹Neural Architecture Search, ¹⁰Automation