AI news We sat down with Kamal Ahluwalia, president of Ikigai Labs, to discuss all things artificial intelligence, including top tips on how to adopt and leverage this technology, and the importance of incorporating ethics into AI design.
Can you tell us a little about Ikigai Labs and how it can help businesses?
Ikigai helps organizations transform sparse, siled enterprise data into predictive, actionable insights using a generative AI platform specifically designed for structured, tabular data. Masu.
The majority of enterprise data is structured, tabular data that resides in systems such as SAP and Salesforce. This data drives planning and forecasting across your business. While there's a lot of excitement about large language models (LLMs) that are ideal for unstructured data like text, Ikigai's patented MIT-developed large graphical models (LGMs) are ideal for structured data. The focus is on solving problems using.
Ikigai's solution focuses specifically on time series datasets, as companies operate based on four main time series: sales, products, employees, and capital/cash. Understanding how these time series come together at key moments, such as launching a new product or entering a new region, is critical to making better decisions that deliver optimal results. is.
How would you describe the current state of generative AI and how do you think it will develop in the future?
The technologies that have captured the imagination, such as LLMs from OpenAI, Anthropic, and others, have a consumer background. They are trained on internet-scale data, and training datasets are only growing, requiring large amounts of computing power and storage. Training for GPT4 cost $100 million, while GP5 is expected to cost $2.5 billion.
This reality works in a consumer setting where costs can be shared across a very large set of users, and some mistakes are just part of the training process. But in a company, mistakes are not tolerated, illusions are not tolerated, and accuracy is paramount. Additionally, the cost of training models on Internet-scale data is far from affordable, and companies utilizing basic models risk having their IP and other sensitive data compromised.
While some companies have gone down the path of building their own technology stacks to use LLM in a secure environment, most organizations lack the talent and resources to build it in-house. doing.
Despite the challenges, companies want the kind of experience an LLM provides. However, even when the data is sparse, the results must be accurate, and there must be a way to avoid introducing sensitive data into the underlying model. It's also important to find ways to reduce total cost of ownership, such as model training and upgrade costs, GPU dependence, and other issues related to governance and data retention. All this leads to a completely different set of solutions than what we currently have.
How can companies create strategies to maximize the benefits of generative AI?
Much has been written about large-scale language models (LLMs) and their potential applications, but many customers are asking, “How do I build differentiation?”
With LLM, nearly everyone has access to the same features, such as chatbot experiences and marketing email and content generation. Even if everyone has the same use case, that's not a differentiator.
The key is to shift your focus from common use cases to finding areas for optimization and understanding specific to your business and situation. For example, if you're in the manufacturing industry and need to move your business outside of China, how do you plan for the uncertainty of logistics, labor, and other factors? Or maybe you want a more environmentally friendly product? If you want to make a , the materials, vendors, and cost structure will change. How do you model this?
These use cases are some of the ways companies are looking to use AI to run and plan their businesses in an uncertain world. Finding the idiosyncrasies and tailoring the technology to your unique needs is probably the best way to find true competitive advantage with AI.
What are the main challenges businesses face when implementing generative AI, and how can these challenges be overcome?
As we listen to our customers, we find that while many companies have experimented with generative AI, only a few have been able to move it into production due to prohibitive costs and security concerns. Ta. But what if you could train your models solely on your own data, run them on the CPU without the need for GPUs, and have the exact results and transparency into how you got them? All the regulatory and compliance issues? What happens when the questions are solved and there are no more questions about where the data comes from or how much data is being retrained? This is what Ikigai has achieved with large graphical models.
One of the challenges we've been helping businesses address is the data problem. Almost 100% of organizations work with limited or incomplete data, and this is often a barrier to doing anything with AI. Companies often talk about cleaning up their data, but in reality, waiting for perfect data can hinder progress. AI solutions that can handle limited and sparse data are essential to enabling businesses to learn from the data they have and account for change management.
Another challenge is how internal teams can collaborate with technology to achieve better outcomes. Particularly in regulated industries, human oversight, validation, and reinforcement learning are required. Adding experts to the loop prevents AI from making decisions on its own, making it important to find solutions that incorporate human expertise.
To what extent do you think a company's culture and mindset needs to change to successfully implement generative AI?
Successful adoption of generative AI requires a significant shift in company culture and mindset, with strong commitment from management and continued education. I saw this firsthand when Eightfold was providing an AI platform to companies in over 140 countries. I always encourage my team to educate executives first about what is possible, how to do it, and how to get there. They need to have the commitment to see it through to the end, and that includes some experimentation and some dedication. They also need to understand the expectations placed on their colleagues so they can be prepared for AI to become a part of their daily lives.
With so much fear-mongering in the press about AI taking jobs away, top-down commitment and communication from executives can go a long way. Executives need to create a sense that while AI won't completely eliminate jobs, everyone's job is everyone's job. In the coming years, things will change for everyone, not just those in the lower and middle classes. Continuing education throughout implementation is key to helping your team learn how to get value from the tool and how to adapt the way they work to incorporate new skill sets.
It's also important to adopt technology that aligns with your company's realities. For example, you need to let go of the idea that you need to have all the data in order to take action. In time series forecasting, by the time it takes him 4 quarters to clean up the data, more data is available and it's probably in a mess. If you keep waiting for perfect data, you won't be able to use it at all. Therefore, we need to be able to learn from the data we have, so an AI solution that can handle limited and sparse data is critical.
Another important aspect is adding experts to the loop. It is a mistake to think that AI is magic. There are many decisions that cannot be made solely by AI, especially in regulated industries. It requires monitoring, validation, and reinforcement learning. This is exactly how consumer solutions have become so great.
Do you have any examples of companies successfully leveraging generative AI?
One interesting example is a Marketplace customer who uses us to streamline their product catalog. They want to know the optimal number of SKUs so they can meet customer needs while reducing inventory carrying costs. Another partner uses us to plan, forecast, and schedule their workforce and balance their hospital, retail, and hospitality workforces. In their case, all their data is stored in different systems, so they need to bring it together in one view to balance employee health and operational excellence. But because we can support a wide range of use cases, we work with clients on everything from predicting product usage to detecting fraud as part of their transition to a consumption-based model.
you We recently launched an AI Ethics Council. Who are the members of this council and what is its purpose?
Our AI Ethics Council aims to ensure that the AI technology we are building is grounded in ethics and responsible design. This is a core part of who we are as a company, and I am humbled and honored to be a part of it alongside such an amazing group of individuals. Our Board of Trustees includes luminaries like Dr. Munter Dahley, founding director of the Institute for Data Systems and Society (IDSS) and professor at MIT. Alam A. Gaboor, associate dean at George Washington University and distinguished scholar of administrative law and national security. Dr. Michael Kearns, Director of the National Center for Computer and Information Science at the University of Pennsylvania. Dr. Michael I. Jordan is a Distinguished Professor in the Departments of Electrical Engineering, Computer Science, and Statistics at the University of California, Berkeley. I am also honored to serve on this Council alongside these esteemed individuals.
The purpose of the AI Ethics Council is to address pressing ethical and security issues affecting the development and use of AI. As AI rapidly becomes central to consumers and businesses across nearly every industry, we believe it is important to prioritize responsible development and the need for ethical considerations cannot be ignored. . The council will meet quarterly to discuss important topics such as AI governance, data minimization, confidentiality, legality, and accuracy. After each meeting, the council will issue recommendations for future actions and next steps for the organization to consider. As part of Ikigai Labs' commitment to ethical AI adoption and innovation, we will implement the action items recommended by the Council.
Ikigai Labs raised $25 million in funding last August. How will this help the development of the company, its products, and ultimately its customers?
We have a strong research and innovation foundation coming from our core team at MIT, so this funding is focused on making our solutions more robust and convening teams to collaborate with our clients and partners. I am.
Although we can solve many problems, we remain focused on solving a few meaningful problems through our Time Series Super App. We know that all companies operate on four timelines, so our goals can include sales forecasts, consumption forecasts, discount forecasts, how to discontinue products, catalog optimization, etc. to cover in detail and quickly. We are excited and looking. We aim to bring his GenAI for tabular data into the hands of as many customers as possible.
Want to learn more about AI and big data from industry leaders? Check out the AI & Big Data Expo in Amsterdam, California, and London. This comprehensive event will be co-located with other major events such as BlockX, Digital Transformation Week, and Cyber Security & Cloud Expo.
Learn about other upcoming enterprise technology events and webinars from TechForge here.
