Recently, Dharmesh, co-founder and CTO of HubSpot, posed an intriguing question during a presentation. When asked "How do you compete with AI?", people interpreted it in two different ways. One-third understood it as 'how to compete against AI', while two-thirds interpreted it as 'how to compete using AI'. This simple survey result carries significant implications, as our perspective on AI could determine our future success.
How AI Works: Beyond Simple Autocomplete
To properly leverage AI, we first need to understand how it works. Breaking down GPT (Generative Pre-trained Transformer) helps us grasp each component's meaning.
Generativetechnology enables AI to go beyond simply searching for information to actually creating content. This is why we call the current era the 'Generative AI era'. While past AI focused mainly on analysis and classification, today's AI can write text, create images, and generate code.
The pre-training process is truly fascinating. Imagine a massive machine with billions of parameters (knobs). This machine learns from all publicly available data on the internet while repeatedly performing the simple task of "predicting the next word." For instance, predicting what word fills the blank in "The horse jumped over the ___". But by repeating this simple task billions of times over months, AI absorbs language patterns and world knowledge.
Some experts dismiss large language models as "advanced autocomplete." While technically correct, this only tells part of the story. This "autocomplete" possesses PhD-level knowledge across all fields.
AI's Limitations and Real-World Challenges
Of course, AI has limitations. First, AI relies only on data from its pre-training period. It's like a PhD graduate in every field who knows nothing about the company's internal systems on their first day. Second, data is frozen at a specific point in time, failing to reflect the latest information. Third, AI sometimes exhibits "hallucination" - not intentionally lying, but confidently providing incorrect answers without knowing they're wrong.
However, these limitations aren't reasons to abandon AI use. In the past, people said to use dictionaries instead of spell checkers, and to avoid the internet because it contained misinformation. Even now, some oppose AI use, saying "life is supposed to be difficult."
Generational Differences in AI Perception
Dharmesh's personal experience is fascinating. He grew up in a small Indian village without phones, TV, or refrigerators. Milk was delivered daily - literally "farm to table" direct delivery. In contrast, his son Sohan learned to swipe before walking, and at 14, uses AI far more creatively than his father.
Sohan dreams of becoming a fantasy novelist and uses ChatGPT for world-building. He creates his own universe with 2,000-word prompts and turns it into role-playing games to test the system. He's essentially commanding computers in what will be the most popular programming language of the future: English.
Telling this 'AI generation' not to use AI is like telling the internet generation to carve letters on stone tablets. It goes against the tide of time.
AI Adoption Strategy in Companies
HubSpot doesn't just encourage its 8,000+ employees to use AI - it expects them to. During hiring, curiosity about AI and willingness to learn are important evaluation criteria. While employees don't need to become AI experts, having curiosity and a learning attitude is essential.
AI adoption isn't a walk in the park - it's like climbing a hill. Resistance is inevitable. The key to success is "dream big but iterate small." Have a grand vision, but start with small experiments and gradually build up.
AI's Future: From Tools to Agents
AI is rapidly evolving. It's transforming from simple text-based interaction systems to action systems that actually perform tasks. As AI gains access to web browsers and internal IT systems, it can handle real work on our behalf.
Progress in code generation is particularly remarkable. If you can express a business or scientific problem in code, AI can likely generate the corresponding software. A new development approach called "vibe coding" has emerged.
The most exciting development is the emergence of 'AI agents' - AI software that performs complex tasks through multiple steps. For example, they can automatically handle composite tasks like "find the CEO's new video, generate a transcript, assess relevance, summarize it, and send it via email."
Human-AI Coexistence: Opportunity, Not Threat
The common answer to "Will AI take your job?" is "Not AI, but people using AI will take your job." But Dharmesh thinks differently. He believes AI will take our jobs but give us much better ones in return. Simple repetitive tasks will decrease, and with digital assistants' help, we'll be able to focus on truly important work.
Humans are more than the sum of the tasks we perform. Getting help from AI doesn't diminish human value. Instead, it allows us to create greater impact and become more valuable.
Future teams will be hybrid forms where humans and AI agents collaborate. This isn't a "you vs AI" scenario, but a "you^AI" exponential amplification relationship.
Practical Advice: Start Right Now
The most important thing is to actually start. Every time you sit at a computer, ask yourself: "How can I use AI for this task?" Set aside skepticism for a moment and try inputting something into ChatGPT or other AI tools. Learning by doing is most effective.
You'll be surprised to discover that AI helps far more often than you'd expect. And there's a beautiful paradox here: as artificial tools advance and AI becomes smarter, we can become more human. Humanity isn't a bug - it's the ultimate feature we all possess.
In conclusion, AI represents exponential opportunity, not existential threat. What matters is shifting our perspective to view AI not as a competitor but as a collaborative partner. By starting with small experiments and gradually building AI utilization skills, we'll be able to focus on more creative and valuable work. Success in the AI era depends not on rejecting technology, but on growing alongside it.