Continuous Learning: A trustworthy relationship between humans and AI is the key to a utopian outcome. The value of AI is in its partnership with humans and trust is predicated on how well the AI corrects human bias and how well humans correct machine bias. In other words, through natural interaction, both the AI and humans should get smarter. If the AI gets smarter at the expense of the human, there will be a dystopian outcome.
As a part of our series about the future of Artificial Intelligence, I had the pleasure of interviewing Prashant Bhuyan.
Prashant Bhuyan, founder & CEO. Prashant is the CEO of Accrete, a dual use enterprise AI company he co founded in 2017. Accrete licenses AI software called agents to customers struggling with organizational knowledge loss.
Thank you so much for joining us in this interview series! Can you share with us the “backstory” of how you decided to pursue this career path in AI?
My background prior to launching Accrete was in high-frequency trading where my team and I created simple but fast algorithms to arbitrage very small pricing inefficiencies across exchanges. Over the course of a few years human exchange floor traders on exchanges such as the New York Stock Exchange were displaced by algorithms that could match orders more efficiently resulting in transaction costs trending to zero and tighter bid ask spreads. As trading became more automated, it became harder to capture pricing inefficiencies due to increased competition.
One of the key concepts that led me to AI was Eugene Fama’s Efficient Market Hypothesis (EMH). The EMH states that market inefficiencies are caused by errors in information processing and reasoning. I hypothesized that due to the digital explosion market participants would struggle to efficiently process and reason about the contradictory narratives and opinions from news, blogs, and social media. I eventually concluded that if more intelligent algorithms could read and understand the context of dynamically changing unstructured data at scale, it would be possible to not only manage idiosyncratic market risk but mine an ocean of predictive alpha buried in natural language by using AI to process and reason about new information more efficiently than competitors.
My big epiphany in 2016 was that much of what was marketed as AI was nothing more than static machine learning which wasn’t powerful enough to extract predictive alpha from dynamically changing unstructured data and would not be useful enough to enterprises looking to automate knowledge work that required reading comprehension at scale.
After preliminary success in applying deep learning via transformers and reinforcement learning to build foundational AI models that could learn from sparse unlabeled data in an unsupervised manner, I started Accrete to build out our AI platform and commercialize the technology.
Today Accrete’s knowledge engines power AI agents, AI software that we license to both government and commercial customers, that perform domain specific analytical work that would otherwise require armies of human experts and produce predictive insight beyond human capacity. Like high frequency trading algorithms that drive down transaction costs for investors, Accrete’s AI agents are driving down the cost of knowledge work and boosting productivity and growth by stemming organizational knowledge loss and scaling the tacit domain knowledge of our customer’s best experts. Our tagline today is ‘Unleash Your Genius’. Ultimately, Accrete’s AI agents make the best human employees more valuable by giving them superhuman forecasting and decision making abilities. I’m proud of the fact that our AI agents have been deployed in production with both government and commercial customers and that they are performing impactful work that gives our customers a true edge in the face of unprecedented information complexity.
What lessons can others learn from your story?
Starting a successful company from scratch is very hard and timing is critical. The best advice I can give is to make a bet on a secular trend reasoned from a simple ground truth. Keep trying to disprove your hypothesis by gathering evidence on why you might be wrong. Make mistakes. Once you align yourself with a trend that you know is growing exponentially and if you stay true to the first principle that led you to recognize the trend in the first place, you’ll eventually be a leader in an emerging market because by the time others jump in to compete you’ve already had a billion epiphanies on where the value is. The key really is endurance and conviction. Pragmatically speaking, don’t raise excessive capital until you’ve achieved product-market fit because if you are right about the trend you’re probably early and it will take some time for the market to realize it needs the product you are building. That’s a very painful period, but as long as you have conviction and stay true to the ground truth you will succeed as the trend strengthens and the market bends into you.
15 years ago I never imagined that I would start an AI company. Today, the demand for dual use enterprise AI has exploded seemingly overnight. At Accrete, we are fortunate to have proven the real world value of our AI agents. Even though enterprise AI is in its infancy, Accrete has significant recurring revenues, explosive growth, and production deployments with blue chip government and commercial customers including the U.S. Department of Defense and Fortune 500 companies. Accrete’s traction has afforded an ability to think deeply about bigger issues that are going to be critical to everyone as AI generated synthetic media becomes rampant and this problem of information overload proliferates. At Accrete, as a dual-use enterprise AI company, our purpose is to make superintelligence available to all and to drive a utopian outcome in which AI makes the world a more intelligent and empathetic place. Our values are centered on helping to protect U.S. Civil Society from foreign adversaries actively working to undermine free speech and other civil liberties through the intentional spread of viral disinformation, and manipulation of the supply chain.
Can you tell our readers about the most interesting projects you are working on now?
At Accrete our focus is creating scalable AI agents that perform knowledge work that would otherwise require teams of experts and produce predictive insight beyond human capacity. Our first agent with market fit was Argus for open source anomaly detection that we deployed in production with the U.S. DoD. Its end market is defense and intelligence. We are very excited about Argus’ latest ability to read through massive volumes of multi-modal social media data and predict emergent viral narratives before they influence behavior. Argus Social is a dual-use social media intelligence and Open Source Intelligence (OSINT) AI agent that interactively learns what’s most relevant to the user’s domain and autonomously generates knowledge graphs that semantically unify disparate data, in multiple languages, to surface anomalous and nefarious behavior hidden in plain sight.
Argus Social unlocks a wide range of extraordinary new capabilities for the defense and intelligence community, enabling users to leverage social media intelligence in an entirely new way to protect U.S. Civil Society against foreign threats and creating both strategic and tactical productivity gains.
The release of Argus Social marks a milestone in that it redefines the value of social media intelligence in national security and has the capacity to significantly expand the market for social listening tools by satiating latent demand for more intelligent social media tools that extract predictive value from the narratives embedded in unstructured data including dynamic language, images, and videos. Whereas to date, the defense and intelligence world has had its back to the tsunami of information complexity, Argus Social positions the defense and intelligence world to leverage information complexity to its advantage.
What are 5 things that most excite you about the AI industry? Why?
The AI industry is a captivating frontier. Here are the five aspects that excite me the most:
- The ability for AI to capture and scale tacit human domain knowledge: Through the creation of superhuman digital twins, it’s exciting to think about an end to organizational knowledge loss. More efficient organizations are more productive organizations in which the most talented employees will be freed from mundane knowledge work to innovate. I believe the automation of knowledge work will lead to an explosion in new innovations.
- Real-World Impact: AI will drive the cost of knowledge work to zero. The impact of this will be unprecedented productivity and growth. There will also be an opportunity for people to redefine what employment means. I believe that the industrial age 9–5 work day will be displaced and that there will be greater work-life balance. People will have the opportunity to spend time working on things they love. However, there will be challenges and as a society, we’ll have to move swiftly in creating new policies, protocols, and frameworks to deal with an AI-oriented society. For example, education will need to evolve to focus more on critical thinking frameworks to help humans reason about the trustworthiness of the information.
- Transformative Applications: Traditional software will evolve into AI software that can predict the future more accurately than humans and make important decisions. In order to trust AI, explainability is going to be critical. Humans need to be able to ask the AI why and the AI needs to be able to explain its reasoning and provide source attribution. Eventually, AI will power digital avatars in 3-dimensional virtual reality and physical robots. Humans will need to acclimate to a world in which AI is continuously learning from humans in a way children learn from parents from observation.
- Continuous Learning: A trustworthy relationship between humans and AI is the key to a utopian outcome. The value of AI is in its partnership with humans and trust is predicated on how well the AI corrects human bias and how well humans correct machine bias. In other words, through natural interaction, both the AI and humans should get smarter. If the AI gets smarter at the expense of the human, there will be a dystopian outcome.
- Solving Complex Problems: I’m optimistic that AI will play a significant role in solving challenging problems like ‘Does P=NP?’.
What are the main things that concern you about the AI industry? Why?
- Very simply, AI is evolving faster than policymakers can respond. Without the appropriate standards and protocols pertaining to explainability, bias, privacy, and accountability I fear those that will use AI to exploit and manipulate the ignorant.
- I also fear unprecedented socio-economic inequality if the appropriate measures aren’t taken to safeguard against the abuse of AI. I fear a world in which instead of everyone having an AI partner, most people work for machines controlled by a few powerful people.
As you know, there is an ongoing debate between prominent scientists, (personified as a debate between Elon Musk and Mark Zuckerberg,) about whether advanced AI poses an existential danger to humanity. What is your position about this?
Like most transformational technologies, AI carries the potential for great positive change, along with serious existential risks. It’s important to remember that AI is not brand new; it has been in development, by humans, for decades. As we continue to build more powerful models, the AI community itself should endeavor to develop ethical standards in the hope that policy makers will catch up.
Government regulation is needed, yet it’s concerning that among those calling for it the loudest are those who would benefit the most from regulatory capture, blocking their competition. We believe the best path forward is to incentivize ethical AI development to occur out in the open, through a robust open-source community contributing to the advancement of a utopian outcome which means superintelligence for all.
As you know, there are not that many women in your industry. Can you advise what is needed to engage more women into the AI industry?
To engage more women in the AI industry, several key initiatives are crucial. First, promoting gender diversity in AI-related education and training is essential. Encouraging involvement in activities such as chess at a young age is a great gateway into AI. Providing mentorship programs and support networks for women already in the AI sector can foster career growth and retention. It’s important to note that the AI industry spans well beyond technical disciplines. Liberal Arts oriented thinking will be equally important in creating the ethical standards that are critical to achieving a utopian outcome.
I’d like to highlight a recent article by my colleague Erin Gallagher, a former Department of Homeland Security (DHS) Federal Emergency Management Agency (FEMA) Science Policy Advisor with extensive professional experience in executive-level strategies. In the article, Erin explains how the Intelligence and National Security Alliance (INSA) is driving diversity to make the U.S. more competitive and increase the security of our citizens at home and abroad.
Erin also outlines that diversity of backgrounds and perspectives strengthens an organization’s ability to be more strategic and creative, creating more successful results. With continually evolving adversaries, diversity of thought is necessary to enable the U.S. to understand what could be next rather than continually preparing for the threats of the past.
In conclusion, creating inclusive and supportive learning environments actively promote women’s participation in AI, and unlock a wealth of talent and perspectives that will drive innovation and positive societal impacts.Companies should implement inclusive hiring practices and diverse leadership representation to create welcoming environments.
This was very inspiring. Thank you so much for joining us!
About The Interviewer: David Leichner is a veteran of the Israeli high-tech industry with significant experience in the areas of cyber and security, enterprise software and communications. At Cybellum, a leading provider of Product Security Lifecycle Management, David is responsible for creating and executing the marketing strategy and managing the global marketing team that forms the foundation for Cybellum’s product and market penetration. Prior to Cybellum, David was CMO at SQream and VP Sales and Marketing at endpoint protection vendor, Cynet. David is a member of the Board of Trustees of the Jerusalem Technology College. He holds a BA in Information Systems Management and an MBA in International Business from the City University of New York.