It was an honor to host Deep Nishar at our recent SV Icons dinner dedicated to Generative AI. Despite facing immense challenges as the first in his family to graduate from high school and college, Deep turned struggles into lessons and setbacks into comebacks. His perseverance and unconventional wisdom led him to remarkable heights, including senior roles at LinkedIn, Google, SoftBank Investment Advisers and now General Catalyst.
Below are some insights from the dinner Chyngyz Dzhumanazarov and I want to share:
1. Tech breakthroughs often result from the combination of various incremental advances over time, rather than a single revolutionary idea. Significant developments in AI and computing, such as convolution neural nets, deep neural networks, and transformer models, have been made possible by the steady improvement in computing power, algorithms, alongside the growth and availability of data and skilled talent.
2. One of the key challenges in AI is the compute requirement for training models and improving them constantly. Quantum computing could potentially address AI's computational demand but is still at lab scale. Not all problems require large models.
3. There will likely be a trend towards 'vertical slicing' of application areas where less compute and infrastructure are required. This approach could enhance efficiency and make AI solutions more accessible to various industries.
4. AI will significantly change our lives, but not in a dystopian, robot-takeover manner. Deep dismisses the extreme AI prediction that robots will kill us all. Instead, AI will continue to augment our lives in a multitude of ways.
5. The shift from deterministic to probabilistic computing is monumental.
6. The complexity of healthcare goes beyond data engineering. Issues such as incompatible data systems, stringent regulatory standards, high operating costs, complex payor-provider relationships and time constraints that limit meaningful patient interactions require holistic solutions.
7. Despite people's growing awareness of data privacy, a gap exists between their concerns and actions as users often trade privacy for convenience or benefits. This necessitates data regulation and embedding responsible innovation into corporate cultures, with a focus on anticipating potential unintended consequences.
8. When starting a company in the GenerativeAI space, avoid addressing derivative problems prematurely. Instead, focus on the core issues until they mature, and then tackle derivative problems.
Deep’s wisdom and perspective were invaluable gifts to our community, thank you Deep Nishar for taking time to attend our dinner!
#generativeai
Organizers: Aizada Marat and Chyngyz Dzhumanazarov
Contributors: David Vernal