Now is the time to celebrate the amazing women leading the field in AI! Nominate an inspiring leader for VentureBeat's Women in AI Awards now before June 18th. Learn more
It is now 2002. You've been lucky enough to get your hands on the first smartphone that lets you send messages to anyone in the world. Your life is changing, right? In the early 2000s, BlackBerry, Nokia, and Ericsson dominated the mobile phone market. In 2007, the launch of the iPhone changed everything and eliminated the previous market leaders.
The iPhone revolution teaches us that during technology hype cycles, the first innovators don't always emerge as long-term winners. In reality, most of the time this is not the case. This is an important consideration for all founders and VCs as the AI hype cycle continues to decline and early-stage generative AI startups remain highly regarded.
What's causing the AI hype?
The debut of OpenAI's ChatGPT sparked tremendous momentum in the AI generation space. Since then, nearly every major technology company has released its own version, and 92% of Fortune 500 companies have adopted the tool. At the same time, a number of “wrapper” startups have emerged offering products based on the ChatGPT model.
One factor that has clearly contributed to this increase is the human tendency to overestimate both short-term and long-term changes. We have already seen predictions about AI replacing jobs being reversed. For example, in 2020, the World Economic Forum predicted that AI would replace 85 million jobs globally by 2025. However, according to the most recent report, AI is expected to be a net job creator.
VB Transform 2024 registration is now open
Join business leaders at a major AI event in San Francisco July 9-11. Connect with your peers, explore the opportunities and challenges of Generative AI, and learn how to integrate AI applications into your industry. Register now
It's undeniable that AI will disrupt the workplace, but pushing the timeline forward only adds to the hype bubble. Again, previous hype cycles show the value of refraining from making such claims. Another example of this is when core neural network research led to breakthroughs in speech recognition and computer vision in the early 2010s.
one article popular science In 2013, it was claimed: “We have to accept that we are probably a lot closer to taking over sentient robots.” This does not diminish the importance of the innovations brought about by deep learning in 2012, but it does mean that we can look to the past to understand today's AI craze. 14 years later, robots may not have taken their place, but the devices we use every day have become more frictionless and more productive.
How to Decide When an AI Startup is Worth Overvaluing
Considering how frothy the AI market is right now, there are a few things to consider when choosing where to place your bets. As with any gold rush moment, it's natural for others to look for pickaxes to build and experiment with things. That means creating horizontal tools and infrastructure solutions.
At the same time, it should be borne in mind that the main difference between previous platform transitions and the current one is the speed of evolution. Established technology companies and startups are simultaneously transforming their technology platforms, and even large technology platform providers are showing remarkable agility in adapting. This means that builds using the Gen AI stack will advance much faster compared to what we saw in the early days of building with the cloud.
If compute and data are the currency of innovation in the AI generation, we must ask ourselves whether startups are positioned to be sustainable compared to established technology companies that have structural advantages and greater access to compute. money required to purchase that access).
The higher up the stack you go, the wider the application opportunities seem. However, where we are in the hype cycle, the reliability of AI output, the regulatory environment, and developments in the state of cybersecurity are key gateway factors that must be addressed for commercial adoption. On a large scale.
Finally, the baseline model achieved performance through pre-training on an internet-scale dataset. What remains to be seen in realizing the benefits of AI is the ability to assemble large, high-quality data sets to build models in more industry-specific domains. It is becoming increasingly clear that the biggest differentiator is not the model itself, but the quality and quantity of data on which it is trained.
Maintaining regulations on radar
Given the excitement and broad potential for innovation in Gen AI and Large-Scale Language Models (LLM), regulators around the world are taking note. Whether it’s President Joe Biden’s recent executive order or EU AI law, startups need to plan for the regulatory landscape.
This doesn't mean they have to know all the answers, but founders should have assessed potential regulatory hurdles and their implications. We're fighting copyright battles and governments are taking positions on what data can and cannot be provided to AI models. More such incidents will unfold.
Understand cybersecurity considerations
Like regulation, AI innovation is outpacing cybersecurity. Businesses need to be aware of when their company data is at risk of exposure due to insecure Gen AI. We've already seen large-scale hacks due to security issues at third-party software providers, which have forced businesses to reevaluate how they vet their vendors. Startups must keep in mind the cybersecurity requirements and reservations of their business.
Gen AI opens up new attack vectors and surface areas for enterprises. From adversarial attacks, instantaneous injection, data poisoning, and model-aligned jailbreaks, many challenges still need to be addressed to make large-scale deployments secure, stable, and robust. AI-infused cyber tools will certainly be part of defense strategies, but protecting AI itself is an emerging subfield in cybersecurity.
AI founders raise a green flag when they demonstrate proactiveness toward regulatory and cybersecurity considerations.
Why data determines the fate of your startup
The biggest factor in determining whether a startup can stand the test of time amidst the noise of the hype cycle is data. For startups to create sustainable value, they need to take control of their data fate. A better question than “What is your Gen AI strategy?” “What is your data strategy?” Because a company's model is only as good as its data. Access to high-quality data draws the line between success and failure. How an organization captures, prepares, and extracts value from its data and how it builds its data flywheel is a critical success factor.
The vast majority of enterprise AI projects stall because companies are unable to leverage and prepare appropriate data sets. Another challenge is that many industrial use cases do not have internet-scale data sets to begin with. At least in some situations, this presents an opportunity for synthetically generated data to proliferate all data an organization has access to.
This is an exciting field for many years and continues to promise breakthroughs that can create feedback loops of synthetic data that improve AI models. We are starting to see notable examples of this at the intersection of autonomous vehicle development, Gen AI, and simulation tools. A similar approach can be seen using a more verticalized foundational model.
Where is the AI hype cycle headed?
It is clear that Gen AI innovation will continue to make waves, and software and APIs will continue to mature in compressed cycles. We will continue to be excited as we see significant performance improvements for models such as Sora, Claude 3, and GPT-5. As with previous hype cycles, we must consider the reality that while nascent technologies may be incredibly promising, they do not provide the whole picture. And we can't jump to conclusions about what the AI generation wave will mean for every industry.
I would argue that the people we need to listen to to understand where the industry is heading are researchers, builders, and doers. Frankly, you don't have to be a better VC to select companies than to predict long-term trends.
Samir Kumar is the co-founder and general partner of Touring Capital..
data decision maker
Welcome to the VentureBeat community!
DataDecisionMakers is a place for professionals, including technical people, who work with data to share data-related insights and innovations.
If you want to read about cutting-edge ideas, latest information, best practices, and the future of data and data technology, join DataDecisionMakers.
You might also consider contributing your own article!
Learn more at DataDecisionMakers