The Twilight of the Chatbots: Why AI Is Getting Good Enough to Do Real Work (and What That Means for Your Business)
If you’ve used ChatGPT, Claude, or any other AI assistant over the past year, you’ve probably noticed they’re getting better. But what you might not realise is just how fast — and how fundamentally — the capabilities are changing.
Ethan Mollick, a professor at the Wharton School and one of the most respected observers of AI progress, recently published an analysis titled “The Twilight of the Chatbots.” His conclusion: we’re moving past the era of simple question-and-answer AI into something far more significant. The latest AI models can now work autonomously for hours on complex tasks that would take human professionals days or weeks.
For Irish small business owners, this isn’t abstract theory. It’s a practical shift that could change how you run your business this year.
The Evidence That Things Have Changed
The numbers are striking. Independent evaluations from METR (a research organisation that measures AI capabilities) and the UK’s AI Security Institute track how much human work an AI can accomplish from a single prompt. These measures are increasing at what Mollick describes as “better than exponential” rates.
In one benchmark, researchers from Epoch found that a recent AI model, working autonomously for 14 hours, built a software package that would take a human engineering team between 2 and 17 weeks to complete. The total cost in AI computing resources: about $251 (roughly €230).
Mollick himself tested a top-tier model that worked for nine hours straight on complex software projects that would have taken a human team well over a week. His experiments suggest that while AI still struggles with many tasks — and can’t replace human judgment in complex situations — the trajectory is clear and steep.
What “The Twilight of the Chatbots” Means for Your Business
The title refers to the idea that we’re moving from thinking of AI as a chatbot — something you type questions to and get answers back — to thinking of it as an autonomous worker that can be assigned tasks and left to complete them.
This shift matters for Irish businesses in several practical ways:
Content creation and marketing. AI can now draft social media posts, write basic website copy, create email campaigns, and even generate images. For a small business without a dedicated marketing team, this is a game-changer. A single business owner can now produce the quality and volume of content that previously required a part-time marketing specialist.
Customer service. The chatbot era is ending not because chatbots were bad, but because they’re being replaced by genuinely intelligent agents that can handle complex customer enquiries, process returns, book appointments, and escalate appropriately — all without a human in the loop.
Administrative work. Drafting proposals, summarising meeting notes, creating invoices, managing email responses — these time-consuming tasks are increasingly within reach of AI tools that cost a fraction of a part-time assistant’s salary.
But Here’s the Catch
Mollick’s analysis also carries an important warning. While these AI systems are improving rapidly, they remain inconsistent. They can produce brilliant work on one task and fail completely on a similar task moments later. They cannot yet be trusted to operate without human oversight on critical business functions.
The practical approach for Irish business owners is what experts call “co-intelligence”: using AI as a capable but imperfect assistant that requires human direction, quality control, and judgment. Let it draft, but you edit. Let it analyse, but you decide. Let it automate the repetitive parts, but you stay involved in what matters.
Practical Steps for Irish Business Owners
Here’s how to start making this shift work for your business right now:
1. Identify your high-volume, low-judgment tasks. Look at what you spend the most time on that doesn’t require your specific expertise. Data entry, first-draft writing, basic research, and email triage are prime candidates.
2. Try multiple AI tools. Don’t commit to one platform. ChatGPT, Claude, and Gemini all have different strengths. The Chinese open-weight models mentioned in Mollick’s analysis are also becoming powerful and cost-effective options.
3. Set up review processes. AI output should never go directly to customers without human review. Build a simple two-step process: AI creates, you approve.
4. Revisit your assumptions quarterly. Given how fast these capabilities are improving, what was impossible for AI six months ago might be routine today. Schedule a quarterly review of what AI can now do for your business.
The twilight of the chatbots isn’t a warning — it’s an opportunity. Irish businesses that learn to work alongside capable AI will find themselves with more time to focus on what truly matters: serving their customers and growing their business.