A Tutor for Every Child; An Assistant for Every Teacher (AI)

Billions of people lack access to the education necessary to thrive and contribute to family and society to their fullest potential. In response, Value Spring Technology is developing an artificial intelligence (AI) tutor, "Ali", to vastly improve the availability of quality education worldwide.  A large body of research shows that each student achieves significantly better learning outcomes in a 1:1 tutoring environment, learning at their own pace and allowing their questions to direct how/what they are taught.  Imagine the solution in each person's hand: a personal tutor and mentor, with unlimited access to information and using best practices to stimulate learning. Students learn what is relevant to them, at their own pace, no matter their location, wealth, or social status, even without books or classrooms. With concurrent advancements in Internet accessibility, students of all ages and economic means, will learn real-world skills through Ali without charge.

For  the reasons described in this challenge, Early Childhood Development was adopted in 2015 as a UN Sustainable Development Goal.  VST will address cognitive stimulation, barrier reduction and learning through personal tutoring delivered by a friendly AI avatar.  200M US children and 650M worldwide children under age 5 are frequently exposed to technology without adult guidance or open-ended, discovery oriented learning, which are critical for achieving the best learning outcomes.  A 2015 "Pediatrics" article described a study of low-income US children showing, "Almost all children (96.6%) used mobile devices, and most started using before age 1. Parents gave children devices when doing house chores (70%), to keep them calm (65%), and at bedtime (29%). At age 2, most children used a device daily... Most 3- and 4-year-olds used devices without help."

According to the “Millions Learning” study by the Brookings Institute, "An estimated “100-year gap” persists between education levels in developed and developing countries, and if business proceeds as usual in the education sector, this gap is not projected to close...250M children around the globe—many of them having spent at least four years of school in a developing country—lack the most basic literacy and numeracy skills."

The goal is to reach the 150-364 Million children who don’t have regular access to education;  girls who are often left out of education opportunities, particularly in developing countries;  those with disabilities that prevent success in many schools; and those whose schools cannot provide more than basic opportunities. In 2017, Peter Diamandis, Executive Chairman of the X Prize Foundation, and author of "Abundance" predicted, “Within the next 20 years, the best possible education on Earth will be delivered by AI for free — and the quality will be the same for the son or daughter of a billionaire as it is for the son or daughter of the poorest parents on the planet." At this point, many cutting-edge education opportunities, particularly ones that involve the emerging AI technology field, are only available to schools and students who can pay for this access, especially since current solutions are so expensive due to their scripted coding. Our team is developing that permanent, scalable solution, at no cost to students.  We are working with experts in the field of AI tutoring, education, and focus groups of teachers in Chicago and Nepal, to ensure that we are developing the right approach and software.

Ali consists of two products based on the same underlying technology:  a tutor for every child to be provided as an app to individuals without charge, and an assistant for every teacher, to be sold into schools.

Value Spring will build and release Ali, an AI tutor utilizing both underlying patented and Watson technology, to provide personalized teaching via personal devices through the internet/Cloud for students with cognitive ages of 2.5 to adult. Value Spring's partners will distribute Ali through existing programs to benefit disadvantaged children everywhere. Ali will guide students through personalized lessons, taking into account each student's cognitive age, learning style and what they have already learned.  Unlike most search engines or Q/A systems with scripted answers to set questions, Ali simulates the way people think and understands the meanings of words, sentences and stories.  Ali responds dynamically in natural language to each student, using Socratic dialogue to stimulate critical thinking.  Ali assesses student understanding, adjusting her approach and response for maximum effectiveness, and learning from each interaction.  

As an assistant for every teacher , Ali will automatically generate lesson plans, course content and Socratic tutorials and perform real-time student assessment, alerting the teacher to learning barriers.  Value Spring software will measure student learning objective attainment and teacher productivity and will create predictive models of student learning deficits, exporting all of this information to multiple existing Learning Management Systems that schools and teachers are already using.

We are partnering with leading worldwide non-profits and teaching institutions which have master, award-winning educators with vast experience in educating students and teachers. With their guidance and input, as well as the utilization of tested tutoring  techniques described above, we are transforming our commercial Artificial Intelligence (AI) meaning engine into a learning system. 

VST's essential innovation is software that understands meaning from natural language.

• VST's AI-based tutor for every student and an AI-based assistant for every teacher, which will increase learning velocity and teacher productivity.  Students will get a personalized tutor and teachers will get collaboration on course material, training and support, based on best-in-class teaching and tutoring practices from Subject Matter Experts' training of the Ali software.  This will free human teachers to do the more complex emotional work requiring human interaction. 
• Today, teaching and preparing lesson plans is a highly manual process, and there is a worldwide shortage of teachers: we will automate this process. 
• We’re using a similar user interface and techniques that have been proven in testing funded by the National Science Foundation at the University of Colorado at Boulder, which showed that even scripted computer-based tutoring produces results at least as good as human tutors.  However, the scripted tutoring used by UCB requires vast programming resources and can only be useful  as far as the scripts are written.  This is true of ALL existing computer-based "natural language" technologies employed to date.

Rather than requiring programmed scripts, VST’s new patents and technology simulates the way people think, understanding meaning in its very essence – the meanings and intent of words, sentences and stories.  This enables a normal conversation with its spontaneous back-and-forth between two individuals.  Ali’s differentiator is its scalable natural language student interaction, including the Socratic method, which no other computer-based system offers.

Ali-Tutor is built on VST's enterpriseMind™ platform.  enterpriseMind is a 9-Dimensional Hypercube of neural networks and digital processes which covers all 9 dimensions of meaning in language. It is this combinatorial processing of language at the word, sentence and story levels which forms enterpriseMind’s natural language understanding. To understand the intent  of two-way conversation, all three levels of meaning must be processed.  In consuming textbooks and other written educational material, Ali-Tutor can learn facts automatically, at small cost, and then scale this knowledge into a 9-dimensional semantic representation for presentation and formative evaluation of student responses.

Kurt VanLehn first introduced the notion of general dialogue management for intelligent tutoring with his characterization of an intelligent tutoring session as an interplay between two control loops; an outer loop which executes once for each new body of lesson content, and an inner loop which executes multiple times per lesson. Each inner loop is one exchange between the student and tutor, typically entailing a presentation of material, or a question, followed by a student response, which then in turn stimulates formative assessment before moving on to the next tutoring move. Multiple inner loops are executed until satisfaction of some completion criterion for completion of the outer loop.  

Based on empirical analysis of tutoring dialogues between real teachers and students, we have discovered a consistent inner loop cadence and tutoring control logic, which ties into VanLehn’s inner/outer loop framework, which will allow successful tutoring.  (See technical papers for more info/references.)

Numerous studies show that education of their children is a primary concern of parents, and all societies engage in ensuring that (at least male) children are educated.  Further, mentally healthy children WANT to learn on their own, as evidenced by studies about self-education in populations where education access have broken down or are unavailable, and through our teams' own experience.  Although our solution in the short/medium term does not address mental health, nutrition, or situational stability for children, in the long term, education does lead to societal improvement in all these areas.  If we succeed in introducing free education for everyone, certain for-profit schools and despots who want to keep their populations (or their females) ignorant may suffer through loss of business or control.  We will be selling our solution into schools to improve their product offerings, and we do hope the despots suffer the consequences of education available anywhere, anytime.  Of course, other roadblocks to success in reaching everyone exist, including availability of internet and access devices in all places and for everyone.  This availability increases daily through decreases in cost of devices and the plans of companies like OneWeb and SpaceX to bring high-speed internet to every place on earth in the early 2020's.  Our goal is to be ready to serve these most remote and most desperate populations with a personal tutor for everyone, focusing on critical thinking skills, before they receive this access and become addicted to video games or believe misinformation presented on the internet.

• Serving now = 0.  Currently we are in pilot stage, and it is important that our solution be well-developed before we start using it to actually help children learn.

• Serving in one year = 100.  It will take more than one year to perfect and test Ali, but we hope to have at least 100 students and teachers involved in testing within this timeframe.  However, once Ali is tested and shown to improve outcomes, we will be able to partner with organizations worldwide for distribution. 
• Serving within 5 years = 100M +.  Once Ali is tested and debugged, there is no reason it cannot be rolled out to the 1.5B potential English speaking students and teachers everywhere, and to people speaking other languages that are already served by our acquired partners.  If Ali performs as we anticipate, we will be able to obtain additional sponsors and to access a great deal of commercial and education customer revenue to support the free distribution and cloud and support services required to roll Ali out as quickly as barriers to rollout can be overcome.
• Serving within 20 years = everyone with a desire to learn.

Within the next year, our goal is to continue to execute our business plan and raise at least $1M to provide fully replicable software  robust enough for demonstration that the technology works in pilot groups overseen by independent evaluation teams. We will also identify and begin to partner with other technology solutions providers to solve those portions of the final solution that others are already working on; we won't waste money/time on creating technology that already exists, such as:

• text-to-speech and speech-to-text recognition, especially for young children, non-native speakers of a language, and people with speech defects.
• visual and auditory emotion sensing and translation of appropriate responses to avatars
• avatar animation of lips, facial expressions, etc, which is also extremely beneficial for engaging young children

During the next 5 years, we will be raising additional funds to integrate this technology, reaching out to content providers and distributors as partners, and actually performing sales and distribution.  These content and distribution partners include:

• textbook publishers for all ages
• Massive Online Open Courseware (MOOC) developers and distributers
• organizations engaged in reaching out to refugee, homeless, and extremely poverty-stricken populations.

Importantly, we will also be selling our Ali teacher assistant solution to school districts and colleges as well as selling other solutions, such as call center automation, made possible by the underlying enterpriseMind software, in order to help raise funds for Ali final development, language re-development, marketing, and distribution.

Barriers include:

1.  Lack of money to get the software developed and tested as quickly as possible, so we can get it into the hands of as many students and teachers are possible.  We need to be able to hire qualified software engineers to help us build it. 

2.  Delay in getting to the right partners because we have limited staff to do the real digging that is necessary to identify the best partners for technology and distribution.

3.  Cultural barriers for girls' education

4.  Legal barriers for technology adoption

1.  Lack of money ...:  We need to be able to hire additional qualified software engineers to help us build Ali, and we are developing relationships with qualified higher learning institutions now to both understand our solution and provide advanced students and graduates to help us finalize it.  We also need money to perform frequent reviews with experts and user focus groups. Extensive testing validates the effectiveness of VST’s enterpriseMind, the platform on which Ali is based. Since enterpriseMind, and therefore Ali, is built on IBM’s Watson platform, IBM has also committed support to ensure smooth delivery.

2.  Delay in getting to the right partners because there are not enough of us to do the real digging that’s necessary to identify the best partners for technology and distribution:  Working with MIT SOLVE would be perfect for partner identification and introduction.

3.  Cultural barriers for girls' education:  Working with existing organizations on the ground in areas where girls and women are purposefully kept from education, and in those areas where girls are under-educated due to financial constraints.

4.  Legal barriers for technology adoption:  Speaking engagements and other outreach to let parents know that we will not sell their children's data.  Money to hire appropriate legal representation when required to roll out our solution in new jurisdictions.  We will also pay attention to local education needs, create flexible adaptation plans, make assurances of student data privacy, and obtain community support, as recommended by experts and our distribution partners.

NA

• Full-time staff: 5
• Part-time staff: 2
• Contractors:  8

The VST team has the technological, managerial, and educational expertise to ensure the success of Ali.

Technical: VST is led by industry-recognized, award-winning software experts who have delivered leading-edge, complex worldwide software solutions for diverse companies including IBM, ADP, United Airlines, AMB Generali and AIG.

Bill Doyle, CEO/CTO, serial entrepreneur, holder of computational linguistics and new technology patents driving VST solutions, NYU AI professor.  PhD Computational Linguistics, SUNY

Dr. Allan Elkowitz, VP Software Leader was  IBM Chief Scientist and Executive Architect, and professor of mathematics and PhD in Physics, Caltech.   

Managerial:  

Taffy Holliday, COO, Serial entrepreneur, mobile software computing pioneer: products used by millions, IBM executive; NYU (math), MBA, Columbia. 

Alice Hoffman, MIT Engineer and NSS/HMP Program Manager for A Tutor for Every Child: successfully managed complex programs of up to $6.5B for 33 years, utilizing well-defined goals, incentives, and risk management techniques to ensure success.  

Educational:

Lynne Zielinski, Education Manager of NSS/SEE: 32 years of physics and space science education experience, numerous national awards for education/technology programs. A former NASA engineer, she founded/managed programs flying student experiments in space.

Frances Dellutri, Deputy Education Manager of NSS: 22 years teaching experience, nominated for the Presidential Award for Excellence in Math and Science Teaching.  She guides teachers and oversees student activities at NSS' International Space Development Conferences, and presented sessions at the Global Hands-On Universe First Asia Regional Conference in Nepal.

• Bill & Melinda Gates Foundation:  Provided grant in 2018 for Innovation in Education.  VST is currently fulfilling the proposal promises, writing a paper for inclusion in a juried journal, and working with Gates to find funding for additional work.
• NSS and its SEE program:  SpacEdge Education (SEE) is a program of the 501(c)3 National Space Society to bring science to students everywhere. Vetted lessons/activities/resources inspire through space science at https://spacedge.academy/ available to all teachers/students.
• IBM Watson re-seller: partnering on commercial insurance solutions. Also working with Watson Education division to integrate the Teacher Assistant with current Watson Education offerings. 
• School of Education at Johns Hopkins University:  evaluating pedagogical approaches used, solution potential and adoption strategy. 
• University of South Florida Computer Science and Engineering Department:  evaluating the scalability/robustness of the VST Ali-Tutor solution.  Discussions underway for integrating their visual/seeing interpretation technology with Ali and co-sponsoring proposals using Ali to teach computer science at universities.

VST's long-term value proposition is aligned with this year's SOLVE Challenge:  to help children of all ages develop the critical learning and cognitive skills they need to reach their full potential, which is especially important for early childhood development, and to help parents and teachers afford the best education for all their children. 

The most pressing global education need is for an autonomous platform that can facilitate the acquisition of knowledge and critical thinking skills as effectively in a Nigerian village as in a USA classroom. Without education – especially education of women – levels of poverty and socioeconomic dislocation will worsen. With education, we can harness the energy of the “youth bulge” for social and economic transformation. We believe that a scalable artificial intelligence focused on intelligent tutoring provides the best hope for a platform that can be effective under suboptimal conditions. Avatars can be adjusted to suit cultural norms. Textbooks written in one language will be read by Ali and their ideas then presented in any language with native speaker assistance. Expert teaching/content can be adapted for parents and village elders in different communities.  We have also seen an explosive increase in wireless and internet connectivity across the third world, enabling easier distribution of intelligent tutoring content than recruitment and placement of a ubiquitous teacher corps.  We are leveraging this wireless internet connectivity to be able to deliver a personal tutor, providing 1:1 Socratic tutoring, in the palm of one’s hand.

VST has the core intellectual property.  We are looking for additional funding from both foundations and capital raise to hire the right skills for software development and subject matter experts in education, and to expand  distribution channels and educational organization affiliations.  We have started in the US, and explored our ideas in Nepal and with teachers from India, Canada and Romania.  We'd like to partner with additional organizations and test our ideas in additional countries, such as the Philippines (in process now).  Once the Ali product is built, we intend to operate the educational offering from  the revenue derived from our commercial solution.

• We are currently seeking sustained donations and grants to continue the development of Ali with the advice of the Gates Foundation.   We need to complete development in order to show Ali to US school administrations and corporations.

• From a revenue-producing perspective, we believe that in addition to the preschool and K-12 school districts that will adopt this solution, there are four distinct opportunity areas:

• Corporate and Government Training -- $141B
• Existing higher educational institutions -- $27B for TA augment/replace
• Text Book publishers  (eBook or hardcopy) -- $14B
• Human tutoring K-12 -- $7B US annually

• The above does not include our broader commercial solution for customer service which has a staggering $1T+ market worldwide.

VST intends to contribute a portion of its commercial Solution revenue towards continuous enhancement of the Ali Tutor software.

• Introduction to potential pilot users and schools for early introduction of solution
• Introduction to potential corporate training partners
• Exposure and introductions to funding organizations and philanthropists that do not accept unsolicited proposals to help us advance and test our solution development.
• Opportunities for free speaking engagements
• Advice on additional fundraising strategies
• Meet potential partners at SOLVE events or obtain introductions thru SOLVE (see next question/answer for specifics)

NA

• The Omidyar Network and the Center on the Developing Child at Harvard: improve our understanding of early childhood development needs and barriers to learning.
• Kahn Academy: discuss improvements to their method of teaching and discuss their facilitation of free distribution of Ali. 
• NetHope:  Distribution of Ali as part of their charitable outreach, using technology for public good.
• Century Intelligent Learning:  obtain their input and thoughts about the value of adding an intelligent tutor and avatar to enhance their existing technology and US reach, understand more about their Institute for Ethical AI in Education and how to expand it
• IBM Watson Education (current partner):  include an integration point with the Teacher Advisor, and resell our Ali-Tutor in their catalogue
• Emerson Collective:  work with Emerson and TeachStrong to understand the most important areas where we can help teachers with our tutoring technology (for them and for their students)
• Hatch Early Learning: determine how we can improve their existing iStartSmart and other pre-school technology programs and whether they have funding available to improve their system.
• Bezos Day One Fund: determine if our solution would help with these local pre-school programs and making them scalable
• Pearson and other large educational publishers: determine their interest in providing ingestible content for Ali to use in tutoring in exchange for access to Ali for their Learning Management Systems.

The AI Innovations Prize may be awarded to teams that leverage artificial intelligence to address the Challenge, which is exactly the focus of VST's work. VST would utilize this prize to advance the development of Ali, The Tutor for Every Child, so that we can get the solution into as many students' hands as possible, as quickly as possible.  We would then seek to partner with The Patrick J. McGovern Foundation and its beneficiary, NetHope.org, to facilitate distribution to refugees and others who need it most.  Schmidt Futures may be interested in getting to know VST and its capabilities and investing in our future, since the natural language processing software underlying our Ali and our Assistant for Every Teacher have a very large market potential, in addition to Ali's capacity to improve the lives of billions of people. VST's existing underlying software, enterpriseMind, can be used with existing technology (not re-developed) to create a leveraged solution to the worldwide teacher shortage and to optimize education worldwide, funded by the revenue created by commercial users of the software.  VST will create independently measurable and verifiable, significant improved learning outcomes starting with small focus groups and expanding as fast as our means, partners and staffing allow.

VST's current staff is more than half women, and we are all committed to providing a quality education to girls, as study after study have proven that investing in girls and women pays back many times in terms of economic development of communities, health outcomes, and education of future generations. In the face of population growth and poverty, universal access to education, especially by women, is one of the strongest forces to be harnessed for increased socioeconomic prosperity, reduced population growth, and social coherence. And yet effective delivery of education to communities without sufficient wealth or stability for schools and teachers remains a daunting challenge We are particularly interested in partnering with local grassroots organizations who promote change in attitudes toward women's education, like TechGirlz, and its counterparts all over the world.  Vodafone's current grantees, Internews and WorldPulse, would be wonderful partners for getting out the news about Ali and promoting its use for women worldwide.  We would use this grant to explore the best partnerships to form in order to include women's education issues in our curriculum development and  to identify additional  information and distribution channels for Ali to reach and positively impact the lives of women and girls.

The income gap in the US between the affluent and under-served has a significant negative impact on education, partly due to inequalities in educational spending starting almost literally from birth.  Research from Rachel Valentino at Stanford University showed that public prekindergarten programs offered minorities and the under-served a lower-quality education.

There are too many inadequate or untrained teachers; outdated, biased or inappropriate curricula and content; lack of teaching materials and technologies; inflexible school schedules; lack of student assessment; and insufficient national education budgets. Many US school districts have this problem.  There are frequently insufficient adequately-skilled teachers with enough resources to provide a quality education for every child, or the ability for a teacher to maintain competency in a quickly changing world. 

Early technology solutions attempting to address these issues have recently met with extreme parent resistance due to perceived privacy issues in some well-publicized cases (with mixed education outcomes) which might turn US parents against all AI solutions that could solve many of these problems.  

VST is committed to student data privacy, and will produce GDPR-compliant software that separates the personal data of students from the big database utilized to improve education outcomes.  Personal data will be encrypted and anonymized before export and will never be used to market anything to the students.  VST would utilize Innospark funds to fully incorporate and determine how to describe and publicize how our solution will not jeopardize their children's privacy in any way, but will lead to improved learning outcomes.