Everyone knows that artificial intelligence is dramatically changing the world around us.
And, although AI is advancing rapidly, it is still facing barriers that must be overcome.
In the lead-up to this week’s impressive NJ AI Summit at Princeton University, here a few areas where innovation will help unlock AI’s full potential:
- AI hallucinates, aka it “makes things up”: This is where AI models generate information it believes is correct, but is in fact misleading, incorrect or entirely fabricated. Hallucinations in AI underscore the importance of ongoing research into understanding and mitigating these errors, as well as the use of Multi Agent Systems, especially as AI becomes more integrated into decision-making processes. Imagine if your calculator just made up the answer. Until we can rely on the output 100% of the time, the uses are limited.
- Energy consumption: At its current adoption rate, we’ll run out of power. Data centers powering AI workflows and training AI models represent escalating energy consumption and a significant area ripe for innovation. As the adoption of AI continues to expand across industries, there’s a pressing need to devise more energy-efficient methods to sustainably power the AI workflows of the future. Without such innovations, the power consumption associated with AI technology is poised to rise substantially, posing challenges to both environmental sustainability and the scalability of AI applications.
- Processing power: LLMs and other AI applications require specific hardware to support their unique processing requirements. With Nvidia as the current market leader, these chips are in short supply and are very expensive. New companies are emerging and existing companies are evolving rapidly to attack this pressing need.
Want to keep up on AI?
- Consider being a member of TechUnited: New Jersey, where we frequently host events on AI like Propelify;
- Follow BetterFutureLabs, which is advancing AI technology alongside folks at Microsoft and others;
- Subscribe to newsletters, including TLDR AI, Towards AI and HackerNews.
- Cognition: AI doesn’t contextualize the world around it. Current AI is observational and reactive, mirroring prior patterns to predict what’s most likely to occur next. This area has been researched for decades with massive strides in innovation more recently making AI more useful. Until AI can understand the world like a human brain, there remain questions on whether it can live up to its full potential.
All this is not to say that New Jersey should not go all-in on AI. Quite the opposite. With our rich history of research, New Jersey has a unique opportunity to help define the future of AI — one that could help elevate the state and our economy with the right collaboration and investment in startups and enterprise-solutions alike.
The challenges here are just a sample of the many complex issues that AI presents and, therefore, the massive opportunity for innovation. As a state with its roots deeply in innovation, the evolution of AI is an opportunity for New Jersey to claim. It is ours to lose.
Aaron Price is the CEO of TechUnited: New Jersey; Justin Trugman is the co-founder and head of Technology at BetterFutureLabs.
