Neuren Media and Analyst Coverage, page-1568

Here’s something that (currently anyway) is a bit unusual.
And it features Neuren!
cheers


https://www.afr. com/wealth/superannuation/can-ai-revolutionise-your-smsf-i-tested-two-top-models-to-find-out-20250314-p5ljl


Opinion
Can AI revolutionise your SMSF? I tested two top models to find out

AI models are still far from flawless as data accuracy and privacy concerns persist AI’s deep research capabilities are staggering.

Tim MackayContributor
Mar 25, 2025 – 5.00am


Using AI for deep research is like quickly scanning shelves with a flashlight in a dark library. Instead of fumbling through endless books, it allows you to filter through thousands of complex books to grab precisely the information you need.

I compared ChatGPT’s Deep Research with xAI’s Grok 3’s DeepSearch. These are AI systems that scour the internet to find, study and summarise information about complex topics.


Deep research is like ChatGPT on steroids. Bethany Rae

Deep research is like ChatGPT on steroids.
When you ask the same basic question of Google or an AI model, they take milliseconds to respond.
But one of my complex prompts [this is the information, sentences, or questions that you enter into a generative AI tool] required 16 minutes of processing time.

It’s been more than two years since I wrote an article about what ChatGPT can do for your investments. Back then I was impressed by AI.

Today, AI models are still far from flawless – data accuracy and privacy concerns persist – but I am staggered by AI’s deep research and time-saving benefits shown in the following prompts:

1. “What information would you need from me to advise whether setting up an SMSF is appropriate? Identify tips and traps I should be aware of in establishing and managing an SMSF.”
This prompt assesses the AIs’ understanding of user needs and superannuation laws.
Both provided comprehensive and useful answers in 15 seconds, but Grok 3 asked nine insightful follow-up questions, leading to a superior response. Win to Grok 3.

2. “Give me (a) An SMSF investment Strategy for my SMSF; (b) the documents required to commence a pension in my SMSF; and (c) A checklist of annual SMSF tasks by month.”
This prompt evaluates their capacity to generate detailed SMSF tools. Grok 3 initially refused to generate legal documents but complied when instructed to assume it could. ChatGPT’s documents, though imperfect, were more than adequate and provided in Word format, saving formatting time. Win to ChatGPT.

3. “This is my current SMSF portfolio. Analyse the investments and explain how I could improve my portfolio.”
This prompt evaluates their analytical skills. I provided a $2 million SMSF portfolio spread across 18 investments including blue chips, term deposits, a property, cash, ETFs and managed funds.
Grok 3’s response was more numerical while ChatGPT’s was more descriptive; a recurring difference between the two AIs.
I was impressed with both excellent responses (concentration risk, more international diversification, more ETFs, etc). Instead of a single answer, both AIs provided a range of practical alternatives, each accompanied by clear pros and cons. Tie.

4. “Give me an investment report for Neuren Pharmaceuticals.”
This prompt tests the AIs’ company analysis capabilities.
I randomly chose a mid-sized company with limited public research.

I trained and worked as an equity sell-side research analyst on Wall Street and in London, so I’m reasonably qualified to assess the output.
ChatGPT gave me a 30-page research report in less than 16 minutes; it would have taken me two weeks. It included an executive summary, price target and recommendation, DCF model, SWOT analysis, ratio analysis, five force analysis, industry analysis, competitor analysis and all financials. Grok 3’s briefer report had more current data due to real-time web access.

Was the research good? Yes, it was very good. Was it all correct? No, both reports contained multiple mistakes. I know this because I fed each report in the other AI to identify any errors and then back into the original AI to improve the report.
As a bonus, the activity scroll allows you to watch the AI reason its way through the various nodes of the problem in real time, which is fascinating. Think of AI as your own research intern; they need ongoing supervision. Checking the accuracy of the AI model’s data myself would only take about a day, which is still much quicker than the two weeks it would normally take to create the report.
ChatGPT assigned a $25 “buy” rating to Neuren Pharmaceuticals (current price $11.03), while Grok 3 initially set a $13 “hold” rating. Upon reviewing ChatGPT’s analysis, Grok 3 identified an error in its DCF calculation and adjusted its target to $16.50 “buy”. Grok 3 considered this a base case, whereas ChatGPT’s outlook was more optimistic. Win to ChatGPT (and I will check Neuren’s price in a year).
The upshot


Overall ChatGPT is the winner. However, a highlight was iteratively pitting two different AIs against each other to analyse and critique the output of the other.
The real winner was using two AIs.
One key lesson is the information you include in your prompt. My final prompt as written above will never give you a 30-page research report. My actual prompt was over 900 words long; effectively the outline of an entire MBA finance course.
AI tips for SMSF investors

1. The more you plan your prompt, the more insanely detailed the AI response.
2. Use meta-prompts by asking the AI to improve your prompts for more precise responses.

3. Use multiple AI tools to leverage their unique strengths and mitigate individual limitations.
4. Ask one AI to validate and improve the other AI’s responses.
5. Check the citations at the end of responses are from reputable sources.