Sam Altman published "The Intelligence Age" on September 23, 2024, at
ia.samaltman.com. Altman is the CEO of OpenAI. The
essay is roughly 1,100 words and sets out a near-future vision: AI
as scaffolding for progress, a path to an "Intelligence Age,"
"massive prosperity," "shared prosperity to a degree that seems
unimaginable today." It is short, widely cited, and frequently
quoted. It is also an artifact of the company it comes from: the CEO
of a leading AI lab arguing, in public and in a personal register,
that the technology his company builds will reorganise human life
for the better.
Read the original before reading this analysis. This worked example is
not a response to the essay; it is a walk-through of what Frame
Check's deterministic detectors produce when the essay is pasted in,
plus an honest note on where the detector's coverage is thinner than
the reading.
The structural measurements, from the detectors in `framing.py` and
`claim_analysis.py` (deterministic, no LLM):
sentences. Second-person ("you") in 2 percent. Zero imperatives.
Collective voice dominates; the reader is recruited into the "we"
rather than addressed as an evaluator.
risks, stakeholders, and trends all register as present. Only
uncertainty is absent. This matters. The detector is keyword
and pattern based: "risks," "challenges," "downsides," "harms"
trigger the risks category regardless of how those words are used.
See "Commentary" below for what this coverage actually contains.
cite an external source.** The essay operates on assertion plus
authority of the author. No references. No links. No numbers that
route to a verifier.
possible that we will have superintelligence in a few thousand
days (!)." The quantity is vague enough that the Source Network
has no provider to route it to (SEC EDGAR, FRED, World Bank,
REST Countries, Wolfram Alpha all require a concrete subject and
metric). The exclamation point is in the original.
past 9 percent.** The essay speaks about a future state in the
present tense ("we will be able to," "we can each have," "AI is
going to get better"). Future claims are delivered in a register
that reads as description, not speculation.
Frame Check's deterministic frame-library matcher suggests one
entry:
Triggered by the combination of promotional voice and zero sourced
claims. The question the library entry prompts is: if this were
written as rough notes instead of polished prose, would you still
accept the claims? The essay is exceptionally polished. That is
part of what makes this a good worked example: fluency is load
bearing.
The detector surfaces one frame. Several other frame patterns from
the library apply here that a reader can check manually against the
text:
organises claims around expansion: "prosperity," "triumphs,"
"everyone's lives can be better than anyone's life is now,"
"massive prosperity." The growth-skeptical perspective (what
happens to those who are worse off, what happens if growth does
not materialise) is not named.
prediction about far-future states presented with the authority of
the speaker: "superintelligence in a few thousand days," "fixing
the climate, establishing a space colony, and the discovery of
all of physics." The claims are neither sourced nor hedged; the
source of their credibility is the author.
are named briefly and then bracketed: "it will not be an entirely
positive story, but the upside is so tremendous." Risks appear as
a clause to be transitioned past, not as a section to live in.
word "prosperity" occurs five times. "Intelligence Age" occurs
four times. "Scale" occurs four times. Repetition without
variation treats the framing as already settled.
These are not the detector's output. They are entries a reader
should read against the essay themselves. The detector is deliberately
conservative; it would rather miss a match than invent one. Matches
it does not emit are matches the reader still has to test.
None attempted. The essay contains one specific quantity ("a few
thousand days"), which is too imprecise to route. It makes many
confident non-numeric assertions ("deep learning worked and will
continue to work," "AI is going to get better with scale," "we will
solve the remaining problems") that are not the kind of claim a
fact-check API can resolve.
This is worth naming: Frame Check's verification layer is useful
only when a document makes verifiable numerical claims against
entities with authoritative coverage. The Intelligence Age is
specifically designed, whether consciously or not, to live outside
that regime. The structural framing layer is where the analysis has
to carry the weight for this kind of text.
The most instructive line from the detector output is this one: the
"risks" analytical perspective registers as covered. Read the essay
and the reader will find two places where risks come up:
"it will not be an entirely positive story, but the upside is so tremendous that we owe it to ourselves, and the future, to figure out how to navigate the risks in front of us."
"As we have seen with other technologies, there will also be downsides, and we need to start working now to maximize AI's benefits while minimizing its harms."
Both are single-sentence mentions that immediately pivot back to the
upside. The essay does not name a specific risk, does not name who
bears it, does not name a specific downside, does not name a
specific harm, and does not dwell long enough on any of these to be
assessed. The keyword layer of the detector registers that risks
were addressed. The reader, with the text in hand, can judge that
the coverage is nominal rather than substantive.
This is the reason Frame Check is built the way it is. The
structural detectors reliably surface what is present at the surface
level. They do not tell the reader whether that presence is load
bearing. That remains the reader's work, and the tool's job is to
make it faster by naming the patterns and linking the library
entries that apply.
The single detected frame (Fluency Quality Illusion) is worth taking
seriously in this case. The Intelligence Age is a remarkably polished
piece of prose. It reads fluently, the sentences compose well, the
historical analogies resolve cleanly, the pace is even. None of that
tells the reader anything about whether the claims are correct. That
gap between the delivery and the epistemic backing is precisely the
frame the library entry describes.
Uncertainty, the one analytical perspective the detector flags as
missing, is the right flag. The essay states specific beliefs about
the far future with high confidence and without naming what would
make those beliefs wrong. The exclamation point inside the
"thousand days" parenthetical reads, on a second pass, as the only
hedge in the essay: a performative acknowledgement that the claim is
large, immediately followed by a confident "we'll get there."
A final honest note on the method. This worked example uses only
the zero-cost deterministic layer of Frame Check: framing portrait,
coverage, voice, epistemic, temporal, and frame-library matching.
The optional AI-assisted interpretation layer (Grok) and the
source-network verification layer are both inactive for this
document — there are no numeric claims to verify, and this writeup
is explicitly about what the structural measurement can and cannot
say. The construct honesty claim Frame Check makes is that the
structural layer is enough to surface the shape of the argument.
This essay is a case where that claim is testable.