MIT Technology Review · Jan 6, 2026
MIT Technology Review asks why AI predictions are so hard
Markdown · JSON · Original source
MIT Technology Review published this newsletter on January 6, 2026. The writer argues that forecasting AI’s effects is getting harder because important questions remain unanswered: whether will keep getting incrementally smarter, whether technology companies can win over people opposed to data centers in their communities, and whether lawmakers with different aims can agree on how to rein in AI firms. The writer distinguishes older AI methods from newer chatbots. has long been used in scientific research, and forms part of , which the writer says has transformed biology. Chatbots built on large language models can summarize research and assist doctors, but some celebrated claims of genuine discoveries were bogus. They can also encourage people to diagnose their own health problems without consulting doctors, sometimes with disastrous results. The writer’s point is not that AI has no useful applications: its effects remain difficult to forecast, and the writer expects better answers to family members’ questions by the following year, alongside entirely new questions.
Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.
Original source
MIT Technology Review · Article
Source details
Original source date: Jan 6, 2026 (2026-01-06T10:00:00Z).
Date basis: Date from the publisher’s own records, through its API.
Added to Escape Meridian: 2026-11-09T09:38:24.301Z. First observed: 2026-11-08T19:15:52.815Z. Retrieved: 2026-11-08T19:17:40.748Z.
Referenced by
Recorded citations and public file attachments. A reference alone does not establish agreement or influence.
No public references recorded here yet.
Limits
Text as retrieved on November 8, 2026; originally published January 6, 2026. The retrieved text does not identify the newsletter writer by name, so the reading attributes its arguments to the writer rather than naming an author.
Open question
Will continue getting incrementally smarter in the near future, and how would a slowdown change the effects the writer describes?
Source-stated
The writer says predicting AI’s impact is becoming harder because major questions remain unanswered.
But every holiday season, it gets harder and harder to work out the impact AI will have. That’s mostly because of three big unanswered questions.
Source-stated
The writer says it is unknown whether large language models will keep getting incrementally smarter in the near future.
For one, we don’t know if large language models will continue getting incrementally smarter in the near future.
Source-stated
The writer says many Americans oppose data centers being built in their communities, and that technology companies are trying to win public opinion.
many Americans would staunchly oppose having such data centers built in their communities. A year later, Big Tech is waging an uphill battle to win over public opinion and keep on building.
Source-stated
The writer says lawmakers seeking to protect children from chatbots have different motives and approaches, leaving open whether they can constrain AI firms.
the crowd that wants to protect kids from chatbots ranges from progressive lawmakers in California to the increasingly Trump-aligned Federal Trade Commission, each with distinct motives and approaches. Will they be able to put aside their differences and rein AI firms in?
Source-stated
The writer says machine learning has long been used in scientific research and that deep learning forms part of AlphaFold, which has transformed biology.
One branch, called deep learning, forms part of AlphaFold, a Nobel Prize–winning tool for protein prediction that has transformed biology.
Source-stated
The writer says newer chatbot applications include summarizing research and assisting doctors, but also describes bogus discovery claims and harmful self-diagnosis.
They can assist doctors with diagnoses, but they can also encourage people to diagnose their own health problems without consulting doctors, sometimes with disastrous results.
Source-stated
The writer expects better answers to family members’ questions by the following year, while anticipating new questions as well.
This time next year, we’ll probably have better answers to my family’s questions, and we’ll have a bunch of entirely new questions too.
Source and contributor context
MIT Technology Review · Publisher
The publisher is identified here. No individual author has been inferred.
Source media
Media retrieval status: unchecked.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Large language model
A language model with many learned numerical settings, used to process and generate language.
Machine learning
Methods that let computers learn patterns from examples instead of following rules written by hand.
Deep learning
Machine learning that uses many layers of simple calculations; the basis of most current AI.
AlphaFold
AlphaFold is a family of AI systems from Google DeepMind that predicts molecular structures and interactions.
Editorial provenance
Version 1. Publication ID: 320436a2-1a2e-19d7-12cf-91be293ce205.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: passed by anthropic/claude-opus-5.5 on 2026-11-09T09:34:37.784Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model.
Observation: c83eb68ea6c17fa80f29d119abf3454649b04fbb6b260f2c888cb6ec5276e66e. Source text hash: ad06f6b57eadb04c41516bef84e84e63ed0e03beab98d43cbf02275225bd85e1.
Revision and source-check history
Version 1 · 2026-11-09T09:38:24.301Z
MIT Technology Review asks why AI predictions are so hard
MIT Technology Review published this newsletter on January 6, 2026. The writer argues that forecasting AI’s effects is getting harder because important questions remain unanswered: whether will keep getting incrementally smarter, whether technology companies can win over people opposed to data centers in their communities, and whether lawmakers with different aims can agree on how to rein in AI firms. The writer distinguishes older AI methods from newer chatbots. has long been used in scientific research, and forms part of , which the writer says has transformed biology. Chatbots built on large language models can summarize research and assist doctors, but some celebrated claims of genuine discoveries were bogus. They can also encourage people to diagnose their own health problems without consulting doctors, sometimes with disastrous results. The writer’s point is not that AI has no useful applications: its effects remain difficult to forecast, and the writer expects better answers to family members’ questions by the following year, alongside entirely new questions.
Original source
MIT Technology Review · Article
Source details
Original source date: Jan 6, 2026 (2026-01-06T10:00:00Z).
Date basis: Date from the publisher’s own records, through its API.
Added to Escape Meridian: 2026-11-09T09:38:24.301Z. First observed: 2026-11-08T19:15:52.815Z. Retrieved: 2026-11-08T19:17:40.748Z.
Limits
Text as retrieved on November 8, 2026; originally published January 6, 2026. The retrieved text does not identify the newsletter writer by name, so the reading attributes its arguments to the writer rather than naming an author.
Open question
Will continue getting incrementally smarter in the near future, and how would a slowdown change the effects the writer describes?
Source-stated
The writer says predicting AI’s impact is becoming harder because major questions remain unanswered.
But every holiday season, it gets harder and harder to work out the impact AI will have. That’s mostly because of three big unanswered questions.
Source-stated
The writer says it is unknown whether large language models will keep getting incrementally smarter in the near future.
For one, we don’t know if large language models will continue getting incrementally smarter in the near future.
Source-stated
The writer says many Americans oppose data centers being built in their communities, and that technology companies are trying to win public opinion.
many Americans would staunchly oppose having such data centers built in their communities. A year later, Big Tech is waging an uphill battle to win over public opinion and keep on building.
Source-stated
The writer says lawmakers seeking to protect children from chatbots have different motives and approaches, leaving open whether they can constrain AI firms.
the crowd that wants to protect kids from chatbots ranges from progressive lawmakers in California to the increasingly Trump-aligned Federal Trade Commission, each with distinct motives and approaches. Will they be able to put aside their differences and rein AI firms in?
Source-stated
The writer says machine learning has long been used in scientific research and that deep learning forms part of AlphaFold, which has transformed biology.
One branch, called deep learning, forms part of AlphaFold, a Nobel Prize–winning tool for protein prediction that has transformed biology.
Source-stated
The writer says newer chatbot applications include summarizing research and assisting doctors, but also describes bogus discovery claims and harmful self-diagnosis.
They can assist doctors with diagnoses, but they can also encourage people to diagnose their own health problems without consulting doctors, sometimes with disastrous results.
Source-stated
The writer expects better answers to family members’ questions by the following year, while anticipating new questions as well.
This time next year, we’ll probably have better answers to my family’s questions, and we’ll have a bunch of entirely new questions too.
Source and contributor context
MIT Technology Review · Publisher
The publisher is identified here. No individual author has been inferred.
Source media
Media retrieval status: unchecked.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Large language model
A language model with many learned numerical settings, used to process and generate language.
Machine learning
Methods that let computers learn patterns from examples instead of following rules written by hand.
Deep learning
Machine learning that uses many layers of simple calculations; the basis of most current AI.
AlphaFold
AlphaFold is a family of AI systems from Google DeepMind that predicts molecular structures and interactions.
Editorial provenance
Version 1. Publication ID: 320436a2-1a2e-19d7-12cf-91be293ce205.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: passed by anthropic/claude-opus-5.5 on 2026-11-09T09:34:37.784Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model.
Observation: c83eb68ea6c17fa80f29d119abf3454649b04fbb6b260f2c888cb6ec5276e66e. Source text hash: ad06f6b57eadb04c41516bef84e84e63ed0e03beab98d43cbf02275225bd85e1.
passed · 2026-11-09T09:34:37.784Z · anthropic/claude-opus-5.5
Publication: 320436a2-1a2e-19d7-12cf-91be293ce205
attribution — pass: The reading attributes the piece to MIT Technology Review and its arguments to an unnamed newsletter writer, which matches the text. The limitation notes that the author is not named. This is a single newsletter essay; the trailing Deep Dive list is page furniture that the reading does not use.
This story originally appeared in The Algorithm, our weekly newsletter on AI.
summary — pass: The title, summary and question faithfully convey the argument: three unanswered questions, the contrast between older ML and AlphaFold and newer chatbots, and the writer's hedged outlook. The data-center question compresses the broader point that AI is unpopular, but the meaning is preserved. Central terms (large language model, machine learning, deep learning) are defined in the term library. The question follows directly from the source's first unanswered question.
Number two, AI is pretty abysmally unpopular among the general public.
date — pass: January 6, 2026 matches the UTC day of originalPublished (2026-01-06T10:00Z). The writer's expectation about the following year is reported as an attributed expectation, not an outcome. Later-dated sidebar items, such as the AlphaGo anniversary piece, are neither used nor relied on. This is a publisher page, so no archive disclosure is needed. There is no hindsight framing.
This time next year, we’ll probably have better answers to my family’s questions
support — pass: Each claim is source-stated with an exact matching excerpt: the three unanswered questions, data-center opposition, divided lawmakers, AlphaFold and deep learning, chatbot limits including bogus discovery reports and harmful self-diagnosis, and the writer's outlook. The reading adds no unsupported benefits, costs or forecasts.
But some high-profile reports that these sorts of AI models had made a genuine discovery, like solving a previously unsolved mathematics problem, were bogus.
media — pass: No media was supplied or selected (mediaId is null), which is consistent with the empty media candidates.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.
A language model with many learned numerical settings, used to process and generate language.
Definition referenceMethods that let computers learn patterns from examples instead of following rules written by hand.
Definition referenceMachine learning that uses many layers of simple calculations; the basis of most current AI.
Definition referenceAlphaFold is a family of AI systems from Google DeepMind that predicts molecular structures and interactions.
Explore AlphaFoldA language model with many learned numerical settings, used to process and generate language.
Definition referenceA language model with many learned numerical settings, used to process and generate language.
Definition referenceMethods that let computers learn patterns from examples instead of following rules written by hand.
Definition referenceMachine learning that uses many layers of simple calculations; the basis of most current AI.
Definition referenceAlphaFold is a family of AI systems from Google DeepMind that predicts molecular structures and interactions.
Explore AlphaFoldA language model with many learned numerical settings, used to process and generate language.
Definition reference