Manish Trivedi
Consumer editor, AI Worth Knowing
22 articles
How It Works
The word was borrowed from human experience and it flatters the mechanism; what actually happens is a very…
· 4 min read
How It Works
Language models do not read letters or words; they read fragments from a fixed vocabulary, and the seams…
· 3 min read
How It Works
The mechanism behind modern language models lets every part of a sequence consult every other part, and its…
· 3 min read
In The World
Generating plausible text became cheap and verifying it did not, and a great many institutions were quietly…
· 3 min read
Limits & Risks
Nothing in the way these systems are built distinguishes a supported statement from an unsupported one, so…
· 4 min read
Limits & Risks
Statistical association is enough to produce behaviour that looks like understanding across most of the…
· 3 min read
Limits & Risks
These models can be asked why they answered as they did and will produce a fluent explanation, which is…
· 4 min read
Limits & Risks
Everything a system knows was fixed at some point in the past, the boundary is blurrier than a single date…
· 3 min read
History
The approach that dominated the field for its first three decades is now treated as a dead end, which…
· 3 min read
Jargon
Continuing to train an existing model on new material is the most commonly recommended and most commonly…
· 3 min read
How It Works
Speech recognition and computer vision were separate disciplines with separate assumptions for decades, and…
· 4 min read
How It Works
The term names a desired outcome rather than a component, and the things actually doing the work sit at…
· 3 min read
How It Works
The step that turned raw predictive models into usable assistants works by learning what people say they…
· 4 min read
Jargon
The technique produces compact models that punch well above their size, and it works for a reason that says…
· 3 min read
How It Works
What gets copied, released or stolen is a set of stored values plus the code that knows how to read them…
· 4 min read
In The World
Systems that learn from examples need examples marked up by people, the work was organised into piecework…
· 3 min read
History
Every task the field has solved was reclassified as ordinary engineering shortly afterwards, which leaves…
· 3 min read
Jargon
The word covers a technical problem about making a system pursue a stated objective and a normative problem…
· 3 min read
In The World
Search engines took material and returned attention, and a system that synthesises an answer keeps the…
· 3 min read
In The World
Occupations are bundles of activities rather than single functions, which is why predictions built on job…
· 3 min read
Limits & Risks
Interfering with training data or with a distributed model file can plant behaviour that lies dormant…
· 4 min read
History
The first mobile robots built a model of the world, planned within it and then acted, and the argument that…
· 3 min read
