Editor’s note: In a report titled 2022 Technology Trends and How They Impact Industry 4.0, Education, and Retail, a team of Orange Silicon Valley analysts with deep subject matter expertise looked at five key innovation trends and how these trends will affect the Industry 4.0, Education, and Retail sectors. This report – from Kevin Baker, David Martin, James Li, and Arpan Soparkar – is now being made available for our Orange Silicon Valley website readers in a series of five weekly articles.
This article examines Data & AI (Artificial Intelligence). Other articles in the series include the Metaverse (including XR/AR/VR); the Future of Work; Sustainability; and Cloud/Edge Computing. Each section of the report includes success stories and promising startups that play into each trend. For additional questions or information, please email David Martin.
Trend 2: Data & AI (Machine Learning & Natural Language Processing)
The Oxford Dictionary defines Artificial Intelligence as “the theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages.” While difficult to describe precisely, what is fundamental to this field of computer science is the ability of machines to learn to solve problems through the analysis of data.
Gartner estimates the AI software market was worth approximately $50B in 2021. Several sources agree this market will see a CAGR greater than 30% during the next six years. This technology is becoming ubiquitous. According to IBM’s Global AI Adoption Index, 74% of surveyed companies are either exploring or deploying AI. Explainable AI is a top priority. 90% of surveyed companies already using AI say the ability to explain how the algorithms arrive at a decision is critical.
Vertical: Industry 4.0 & AI
Industrial environments are filled with data that can be leveraged to solve problems and drive efficiencies in industrial processes. Once data is harvested with sensors, artificial intelligence is the tool of choice to use the data to provide actionable insights and automation. From predictive digital twins of manufacturing tools to computer vision in self-driving robots in fulfillment centers, the versatility of AI enables an endless variety of use cases for industrial applications.
Example
Micron Technology, the memory chip manufacturer, sees no bounds to the use cases of AI to see, hear, and feel in its factories. The company has an extremely sensitive manufacturing process involving as many as 1,500 individual steps. Micron uses computer vision for defect classification, acoustic listening for preventative maintenance, and thermal imaging to proactively detect temperature variations. The company claims the introduction of various forms of AI has resulted in a 25% faster time to yield maturity, a 10% increase in manufacturing output, and 35% fewer quality events.
Startups
Instrumental, started by two ex-Apple mechanical engineers, claims that 20% of every dollar spent in manufacturing is wasted in the $8 trillion industry. The startup offers a cloud-based platform delivering AI-powered proactive defect discovery, end-to-end failure analysis tools, and remote real-time build monitoring.
Vertical: Education and AI
“Our intelligence is what makes us human, and AI is an extension of that quality,” says Yann LeCun, the Director of AI Research at Facebook and a professor at New York University. With the vast and deep proliferation of big data has come a rapid reduction in data storage costs and an exponential rise in data analytics, which has transformed many sectors, including education. Capturing multiple data points about how students learn can help train AI algorithms to address teacher shortages effectively, adapt examination methodologies, and improve pedagogical techniques, enhancing customized learning and student evaluations. Between 2021 and 2024, with a CAGR of 29%, the global personalized learning market will surpass $2B. However, especially in K-12 education, ethical data collection and usage applications will continue to be a challenge.
Example
Based in Pittsburg, PA, Duolingo is probably a poster child success story of data-driven learning. Founded in 2011, the firm raised just over $180M in various rounds until it became a public company in July 2011, at $4B market capitalization. The most downloaded education app in the history of the App Store, it became a de-facto tool for over 1.2B people to learn a new language. Leaning heavily on captured data about how people learn, the app has improved while helping people. It boasted a next-day retention rate of 55% in 2020, up from 13% in 2012, which shows how these data points helped engage its customer base.
Startups
Based in Berkeley, California, Myntor is an early stage ed-tech startup that builds online courses which answer student questions in real-time, using conversational AI, starting with standardized tests such as the ACT and SAT (US standardized tests for university admissions). This crucial feature helps students strengthen their understanding of concepts in a scalable way. Much faster than their counterparts in traditional schools, the students outperform national average pass rates set by live teachers. Myntor raised a pre-seed round at a $2.5M pre-money valuation.
Even though 80% of 6.1M community college students across the US want to transfer to a 4-year university and 68% of universities fail to meet their annual enrollment targets, the process is unnecessarily bureaucratic and cumbersome. Edvisorly, another Berkeley, CA-based startup, takes a data-driven approach, connects these dots among all stakeholders, helping students navigate this process for free while minimizing unnecessary credits and transfer time.
Vertical: Retail & AI
AI continues to redefine retail both online and offline from cashier-less checkout and inventory management to price optimization, customer service, and even call center management.
Use Cases
- Cashierless checkout & inventory management: companies like Amazon and startups are testing and pushing cashierless solutions into retail. Solutions vary from custom wide-angle camera solutions with weight-sensitive plates under products to off-the-shelf cameras that recognize people and objects for purchase as they interact and eventually leave stores. Roaming robots and cameras are also used to track and maintain inventories automatically.
- Visual search: Google images, websites, and many eCommerce apps now enable consumers to search using photos to find and price the products they want to buy.
- Customer service: Chatbots drive messaging services that start their life by routing customers to the agents empowered to solve their problems and mature into bots that can answer and solve customers’ issues directly and without a human in the loop.
- Price predictions: Systems evaluate many inputs from the weather, the day of the week/month/year, and pricing will affect sales, revenue, and profits.
- Voice search: Consumers can use voice to buy products from Amazon Alexa and Google Home, although these purchases are rare.
- Virtual fitting rooms / try before you buy: Consumers can scan their body to virtually “try on” clothes or use an augmented reality app to try on shoes, sunglasses, makeup, and more.
- Call center management: Systems can coach call center workers through voice monitoring software that encourages them to increase their energy and identifies dissatisfied customers
Example
Herschel Supply Co., a Canadian company selling hipster retro backpacks and accessories, worked with Vertebrae to implement 3D & AR product visualizations on their website. For shoppers that engaged with the 3D and AR experiences, the conversion was up 153 percent. The overall conversion rate increased by 23% in both units and revenue. (Source)
Startups
Standard Cognition, Grabango, Zippin, and Radar offer cashierless checkout solutions like Amazon Go.
OTO Systems develops a next-generation language processing voice technology to unlock valuable data from voice interactions. OTO was acquired by Unity in August 2021.
Avaamo is a deep learning Virtual Assistant Platform for the enterprise.