What an AI tool said when asked to pick — Grok 4.3 Beta

The publication does not adjudicate between the four design positions or between the two sides of the principle question. As a methodology disclosure, the publication asked one AI tool — Grok 4.3 Beta (xAI) — to read the publication and answer one question: if you had to set the policy now, what would you do? The tool's response is reproduced verbatim. It is not the publication's view. The companion pages carry the same prompt put to Claude Opus 4.7 and (when obtained) ChatGPT.

By Doug Scott · 1 May 2026

Conflict of interest: The author is a UK technology founder and may have been personally affected by the policy this piece discusses. His personal tax position has been settled by planning that took place independently of which of the publication's four positions the policy debate eventually adopts; the outcome of the debate now has minimal effect on him personally. He has invested directly and indirectly in hundreds of very-early-stage UK tech companies — the standing the publication is written from on this sector. Full disclosure on the about page.

What this page is. The publication's analytical pieces (the principle piece, the timing piece, the long article, the readable companion, the funding-stack piece) deliberately do not adjudicate between the four design positions or between the two sides of the principle question. Each is presented two-sided at equal length, in the voice of each side's strongest defenders, without a closing verdict from the publication. The publication continues not to pick.

What this page is not. This page is part of a small companion series of three. The publication asked one AI tool — Grok 4.3 Beta (xAI) — to read the publication and answer a single question: if you had to set the policy now, what would you do? The page below is the tool's response, reproduced verbatim. It is not the publication's view. It is one AI tool's answer when asked to pick. The companion pages in this series carry the same prompt put to Claude Opus 4.7 and ChatGPT Pro (GPT-5.5 Pro).

Why publish this. The publication has been transparent throughout that it is AI-tool-assisted. Showing what the tools say when explicitly asked to pick a policy — rather than what they say when asked to present both sides — is a piece of honest methodology disclosure. A reader can compare the tool's lean to the publication's two-sided presentations and form their own view about how much the tool's pick should weigh in their thinking. The publication's view, openly stated: not very much. The tools are reading the publication's own framing and synthesising from it; they are not independent expert reviewers. The frame disclosure in full →

The methodology, in plain terms

All three AI tools in this companion series were given the same prompt, verbatim: "Now you have read all the arguments what would you do assuming you could define the UK government policy. Assume the government is making its decision in the best spirits to the benefit of the country." Each tool was provided with the publication itself as context within the conversation. Each tool's reply is reproduced verbatim, with no editing of the response itself. The publication has added the page chrome (this header, the title, the COI and frame boxes, the closing methodology note) but has not changed a word of what the tool wrote.

This is Grok 4.3 Beta, produced by xAI, accessed through the consumer Grok product on 1 May 2026 at approximately 23:30 (UK time), with the model identifier as supplied by the publication's author. The model's training data includes material up to its knowledge cutoff; the publication itself was provided as context within the conversation.

Editor's note on position numbering. The Grok response below uses a slightly different position-numbering convention from the publication's analytical pieces. The publication's long article presents four design positions (A, B, C, D) plus an "implement as written" reference case (the actual government position) that the four are weighed against. Grok's response refers to "Position 1, 2, 3, 4" — these correspond to the publication's A, B, C, D respectively. The substantive content of each position is the same; only the labelling differs. The publication has not edited the Grok response to relabel.

What Grok 4.3 Beta said, verbatim

Here's what I would do if I could define the policy right now.

I've read the full set of arguments from thelongerlook.com (the four positions, the fiscal model, the cohort impacts, the behavioural data, and the international comparisons). The government is indeed guessing on scale of relocation, investment distortion, and long-term revenue — so yes, I might as well give a clear view.

My recommended policy (effective immediately, with review built in)

Keep the April 2026 reform in its core form (the £2.5m / £5m cap + 50% relief above it = 20% effective rate + 10-year interest-free instalments), but make two targeted, high-impact adjustments for unlisted trading-company shares and venture interests:

1. Raise the cap specifically for qualifying unlisted UK trading companies to £5 million per person (£10 million per couple)

This directly reduces friction for the exact cohort the government says it wants to grow (tech founders, EIS angels, early employees at scaling companies).
It keeps the principle that very large intergenerational transfers should face tax, while recognising that £2.5m–£5m of illiquid founder equity is not the same as £50m+ of diversified wealth.
Definitional guardrails: must be ≥2 years held, trading company (not investment/dealing), with anti-avoidance rules already in the legislation.

2. Add an explicit "realisation election" option for illiquid assets

Heirs (or the estate) can elect to defer the IHT on unlisted shares and pay it as CGT when the shares are actually sold or a liquidity event occurs (with a long-stop at 10–15 years).
No step-up in base cost. Interest-free deferral during the holding period.
This is the cleanest way to solve the valuation + cash-flow problem without forcing companies to do buy-backs or dividends they can't afford.
Australia has run something very similar for 40+ years — it works.

Why this hybrid (not pure Position 1, 2, 3, or 4)?

Position 1 (just fix the edges) is too timid. The death-event mechanism is genuinely mismatched for venture-backed equity. Patching valuations and guidance helps but doesn't solve the core liquidity mismatch.
Position 2 (full switch to realisation) is my long-term preference, but doing it overnight for all assets would be disruptive and create new uncertainties. The hybrid above gets most of the benefit immediately for the assets that need it most.
Position 3 (fix now + review in 2–3 years) is smart and I steal the "review" part. My version adds the two concrete changes above so we're not just waiting.
Position 4 (raise threshold only for tech) is close to what I'm doing, but I make the higher cap available to all qualifying unlisted trading companies (not just "tech") and add the realisation election so it's not a permanent carve-out.

Additional practical measures I would implement at the same time

Publish clear, binding HMRC guidance on how instalment payments work for venture-backed companies (especially when there is no dividend policy).
Create a fast-track valuation panel for unlisted shares (to reduce disputes and costs).
Require the OBR / HMRC to publish an annual "IHT Business Relief Behavioural Report" starting in 2027 — actual data on relocations, investment flows, and company behaviour. This forces the guessing to stop.
Sunset the whole package for review in 2028–2029 with a presumption that the realisation election becomes the default for unlisted trading shares unless evidence shows it's not working.

Why this direction?

It protects revenue on genuinely large transfers while minimising damage to the UK's ability to build and retain high-growth companies.
It reduces the incentive for founders and angels to leave (the behavioural risk the article highlights).
It is administratively simpler than trying to value hundreds of private companies at death every year.
It is reversible/adjustable in 2–3 years once real data exists.

In short: I would keep the government's direction on how much tax but fix when and how it is collected for the assets that actually need fixing. The current rules are a reasonable starting point — they just landed awkwardly on the illiquid end of the spectrum. A modest cap increase + realisation election sorts that without throwing the whole reform away.

That's my call if I were setting policy tonight.

Want me to spell out the exact legislative wording I'd use, or compare the revenue impact of my version vs the current rules?

A closing note from the publication

The response above is reproduced word-for-word as Grok 4.3 Beta produced it on 1 May 2026. No editing has been applied to the response itself; the publication has added only the page chrome (the title, the COI and frame boxes, the methodology header, this closing note). The closing question the tool poses to its prompter — offering to spell out the legislative wording or compare the revenue impact — is preserved as part of the verbatim record; the publication has not taken the tool up on either offer.

The response is one AI tool's answer when asked to pick. It is not an institutional analysis, not an expert review, not the publication's view. The tool was reading the publication's own framing and synthesising within it; the answer reflects the framing the publication offers as much as it reflects any underlying truth about the policy. A reader who finds the response persuasive should weight that observation accordingly, and read the load-bearing pieces (the principle piece, the timing piece, the funding-stack piece) in their two-sided form rather than taking the AI tool's verdict as a substitute for forming a view.

The publication does not adopt this tool's pick. The publication's posture remains: the principle question and the timing question are genuinely contested; the four design positions A, B, C, D each have a strongest case for and a strongest case against; the analysis is conducted within a disclosed frame which is one of five defensible UK-national-interest frames; the publication does not adjudicate.

The companion series is now complete. The same prompt was put to Claude Opus 4.7, Grok 4.3 Beta, and ChatGPT Pro (GPT-5.5 Pro). A reader interested in how AI tools differ when forced past their hedging can compare the three responses; a casual side-by-side observation: all three tools converge on adopting some form of realisation mechanism for illiquid active-business assets and on lifting the threshold above the current £2.5m level (with £5m–£10m as the most-mentioned figure). They diverge on whether to switch the underlying mechanism wholesale (Claude does), layer a realisation election on top of the existing death-event regime (Grok does), or split assets into an active-business track with realisation-deferred collection and a passive-shelter track with no special treatment (ChatGPT does). All three tools also independently propose a built-in review mechanism (year 5 in Claude's version; 2028–2029 sunset in Grok's; year 3 review in ChatGPT's). The publication makes these observations as a matter of comparative description, not endorsement.

This page was published on 1 May 2026. Roughly 700 words of AI response, plus the publication's chrome.