What AI Will (& Won’t) Do: AI Predictions Dinner Takeaways
We hosted our inaugural AI Predictions dinner with a curated gathering of thought leaders – corporate innovation officers, founders, journalists, and researchers – where we discussed 8 predictions crowdsourced in advance from the participants. This continues our work in partnering with founders through our AI Start fund, educating CXOs through our network, and furthers our POV on AI as a force for societal good.
Here were some of the key takeaways that came out of our debate:
Prediction: AI Will Redefine What It Means to Be Human
Even today, we’re not clear on what it means to be human;
AI will make us more efficient and effective, which tech has always done. This may free us up to better uncover our potential as a species (tech exploration, more time with friends and family, more fulfilling lives, etc.).
Prediction: AI Will Create More Jobs Than It Eliminates
AI will need more guardrails than humans in order to be fully effective;
Outsourced jobs are the most at-risk, but we believe employees will elevate themselves to get above the waterline (e.g. Medical transcribers will become medical coders);
AI tends to lift the bottom: Bad employees become mediocre employees, and mediocre employees become more productive. This frees up top contributors to become more strategic.
Prediction: Responsible AI Will Grow from a Feel-Good Idea to a Shared Reality
Unfortunately, shared reality is an elusive idea, so defining the right parameters for what constitutes responsible AI will be difficult;
Legislation will be the closest stand-in to setting the floor for boundaries.
Prediction: The AI Investing Valuation Bubble Will Burst with a Bang Not a Whimper
We are still in the early stages of the hype cycle, there will still be more momentum in the space before the dip, but investors will converge on fewer and fewer companies as time goes on. It will take 3-4 years to reach the end, and then markets will likely overcorrect;
GPU, cloud & foundational models are driving valuations of AI startups super high;
We still don’t know what a great company looks like in the AI application space.
Prediction: Big Companies Will Win Vs. Most Startups
Very mixed opinions here: Large companies may have enough capacity, moat, and large shoulders for acquisitions to be the big winners here, but the startup community will move faster, exercise more agility, and attempt to dislocate the current leaders.
Prediction: NVIDIA’s Supremacy Will Be Challenged
Yes – incumbents always face competition;
In this case, the cloud providers are best suited to compete, as they have access to workloads in the training space;
However, NVIDIA can go vertical and up the stack with better cloud service to compete.
Prediction: Open-Source Models Will Win Over Closed-Source Models
In the enterprise space, on a short time horizon, closed-source will win out. It has a head start, is convenient, is safer/more secure, and just works better;
In the enterprise space, on a long time horizon, open-source will win out. It’s more flexible, cost effective, and the security + tooling will continue to improve over time.
In the consumer space, on both a short and long time horizon, closed-source will win out as Google and Apple own the ecosystem.
Prediction: Autopilots Will Be More Prevalent than Co-Pilots
Autopilots will ultimately prevail over co-pilots, but co-pilots are the initial gateway;
This transition will be enabled as trust in these models improves, which will require time and the right context (some situations have more regulation, or require more trust than others).
We also asked participants to name 1 task that AI will not automate and got a varied and entertaining range of answers including: plumbing/plumbers, parenting, cooking/eating, playing sports, love/feelings, and journalism!
