AI and Technology

ChatGPT-5 ‘doesn’t live up to expectations’ but ‘hallucinates less’

👓 30 second read

 

Did you notice that ChatGPT 5 quietly became available in your free account last week? 

 

  • FREE users can access GPT-5 with usage capped to ‘one thinking message per day’

  • PLUS subscribers get 160 messages per three hours (fallback to mini afterwards)

  • PRO and Enterprise users get unlimited access

 

OpenAI’s eagerly awaited latest upgrade has been met with a modest response but is claimed to ‘hallucinate less’ than o3, which First National has caught out completely fabricating statistics from clean data provided in the form of the Cotality Home Value Index.

 

OpenAI has also quietly admitted AI has hit its ceiling. It's not getting exponentially smarter anymore.

 

However, GPT-5 seems faster and better in practice, although frustratingly weak at remembering rules you may have set around image creation.

 

Google AI search splits: What should you do?

👓 3.5 minute read

 

 

The real estate industry is facing a fundamental change in how Google delivers answers. The introduction of two distinct AI-powered search formats, AI Mode and AI Overviews, marks a shift from a single set of ranking rules to a dual-path system with very different behaviours.

 

Your agency’s visibility, citation density and content volatility differ significantly between Google’s two AI search modes. For digital marketers, understanding this split is essential to sustaining visibility in an increasingly AI-led search landscape.

 

What Are AI Mode and AI Overviews?

 

Google’s AI search experience now presents results in two separate ways, depending on the query, user intent and search context.

 

  • AI Mode is a full-page, conversational interface that delivers summarised results in response to user queries. It functions like an interactive assistant and surfaces responses drawn from a curated set of sources. These results often appear in Google’s Search Generative Experience (SGE), especially when users opt in to a more exploratory or research-style journey.

  • AI Overviews are compact summaries that appear within traditional Google search results. They sit at the top of the results page and aim to provide a quick, AI-generated answer to the user’s question. These overviews are often densely cited and are designed to summarise a topic before the user scrolls to organic listings.

 

In simple terms, AI Mode acts as a full AI assistant. AI Overviews serve as enhanced snippets layered on top of classic search results.

 

There are important distinctions between the two modes:

 

1. Brand Visibility

 

AI Mode is significantly more brand friendly. Brands are mentioned in 90% of responses, and the system surfaces nearly four times more unique brands than AI Overviews.

 

In contrast, AI Overviews are much more selective. Brands are only mentioned in 43% of responses, meaning many organisations are entirely absent from this layer of search.

 

2. Citation Patterns

 

AI Mode uses a small number of summarised source cards, typically between 5 and 7, that appear below the main answer. These give users optional links to learn more.

 

AI Overviews, however, are citation heavy. A single overview may contain 20 or more inline citations embedded directly in the text. This signals a much stronger focus on authoritative sourcing and suggests a higher bar for inclusion.

 

3. Volatility and Stability

 

AI Overviews are highly dynamic. Content shown in Overviews changes up to 30 times more week-on-week compared to AI Mode.

 

AI Mode is far more stable, favouring a consistent set of sources and formats. For marketers, this means AI Mode is better for long-term visibility, while AI Overviews may require frequent updates and monitoring.

 

Two modes, two strategies

 

The split is not just a technical detail. It reflects a fundamental shift in how Google interprets search intent and surfaces content.

 

  • AI Mode acts as a broad discovery engine. It casts a wide net and allows more brands to appear. Entry barriers are lower, and content inclusion is relatively predictable.

  • AI Overviews are highly curated and behave more like editorial digests. They are selective, volatile and more likely to reward fresh, high-authority content.

 

What should you do next?

 

This dual-mode system calls for a more nuanced approach to content development and SEO strategy. Here are four practical actions to take:

 

  1. Build for Scale and Breadth

 

AI Mode rewards broad coverage. Estate agencies should develop content that addresses a wide range of user needs across different stages of the customer journey. FAQs, guides, explainer content and evergreen articles are ideal for this format.

 

  1. Raise the Authority Bar

 

To perform in AI Overviews, your content must be highly credible. That means citing trusted sources, earning high-quality backlinks, and demonstrating expertise. Google is clearly favouring content that meets high editorial standards in this mode.

 

  1. Monitor Volatility and Adapt

 

The high turnover in AI Overviews means visibility can shift rapidly. Use tracking tools to monitor when your content is included and why it might drop out. Treat Overviews as a real-time feedback loop for search algorithm changes.

 

  1. Invest in Structured Data and Content Design

 

Both modes rely heavily on structured information. Schema markup, clear formatting and consistent headings help AI systems understand and summarise your content. Well-structured pages are more likely to be included and cited.

 

Search in 2025 is not just about ranking. It is about presence, visibility and adaptability across multiple layers of AI. Brands that understand the structure and behaviour of both AI Mode and AI Overviews will be best placed to lead in this new era of search.

 

Try our new prompt for a blog post that responds to your website’s new SEO needs. You’ll find it in the Marketing Prompts section of our new AI Prompt Library. (Links Below)

 

Prompt Library to speed AI uptake across First National

👓 30 second read

Start exploring First National's Prompt Library today
Start exploring First National's Prompt Library today

 

The best way to boost productivity using AI in your business is to just get started. Choosing a Large Language Model (LLM) such as ChatGPT is a great place to start. Just remember to follow our guidelines on how to keep your data as safe as possible.

 

While there are simple guidelines about how to engineer a prompt effectively, the process of learning can be accelerated by now deploying and refining some of the prompts now available in First National’s Prompt Library. Once you get started, you can refine the prompts so they’re best suited to the way your office works. 

 

Start exploring our Prompt Library today and check back for periodic updates. There are new additions to explore. Let’s share our experiences via The Pitch Facebook group.

 

 

Individuals using AI exceed humans working in teams

👓 1.5 minute read

 

Research on 776 experienced people working on product innovation challenges at Procter and Gamble has revealed how Generative AI is fundamentally reshaping teamwork and expertise. The findings show how critical it is for agents working in small real estate agencies to seize hold of AI’s benefits – even at the most basic level.

 

A working paper called the Cybernetic Teammate has compellingly shown:

 

  • Traditional human teams without AI produced noticeably better results than individuals working alone

  • Individuals working with AI-produced solutions at a quality comparable to or better than human teams working without AI

  • AI cuts down the time spent on tasks, with individuals using AI performing 16.4% faster than those without, and teams with AI working 12.7% faster

 

So, AI offers a staggering leap in capability for one person. However, the study also revealed something deeper psychologically - AI users were less confident in the quality of their work outputs. This perhaps highlights the human role in guiding and refining what the AI produces – not just accepting it without challenge or fact-checking.

 

Also revealed was:

 

  • AI appeared to break down the barriers or silos between different business departments

  • Individuals working with AI produced more balanced solutions

  • Users tripled their chance of creating real breakthroughs in their work

 

The best results were shown to be achieved when users ‘sculpted the output’. Rather than accepting the AI solutions, users who continued to prompt (or chat) with their LLM – sometimes with an additional 19 to 24 prompts, like having a conversation – produced the most balanced and nuanced results.

 

AI is no longer a passive tool. It is a cybernetic teammate that can amplify the quality of your work, help you achieve breakthroughs, and lift your efficiency.