Brains and AIMiddleLesson 3026 min read

What AI can learn from brains

Brains have inspired AI before, and may again. Learn which ideas are being borrowed, what a wiring diagram alone does not tell you, and the ethical questions around brain data.

✦ Ask the tutor🧪 Lab · 30 min📝 Lesson test
Lesson in motion

In 60 seconds

What AI can learn from brains

Brains have inspired AI before, and may again. Learn which ideas are being borrowed, what a wiring diagram alone does not tell you, and the ethical questions around brain data.

1/5
In simple words
Birds inspired people to build aeroplanes. But aeroplanes do not flap their wings. People learned the idea that mattered, the shape of the wing that lifts, and built it their own way. AI and brains are similar. AI can borrow good ideas from brains without copying every detail.

Ideas AI has already borrowed

  • Neurons and weights: the first artificial neurons were inspired by real ones.
  • Layers for vision: early image networks took inspiration from studies of how the visual part of the brain processes simple edges first, then more complex shapes.
  • Learning from reward: ideas in reinforcement learning are similar to reward signals seen in some brain cells.

Ideas being explored now

Brain ideaWhat AI might gain
Sparse, event-driven activityMuch lower energy use (lessons 6 and 7)
Learning continuously without forgettingModels that learn new things without losing old skills
Learning from few examplesLess data needed for new tasks
Tight loop between body and brainBetter robots that sense and act together (lesson 4)
Wiring patterns from connectomesNew ideas for network structure, tested by experiment

What a wiring diagram alone does not tell you

The FlyWire map is a huge achievement. But scientists are careful about what it can and cannot say on its own:
  • Activity: the map shows possible routes, not which signals are flowing at a given moment.
  • Exact strength and sign: synapse counts hint at strength, and whether a connection excites or inhibits must be measured or estimated.
  • Chemicals that change the mood: some chemicals, called neuromodulators, can change how whole groups of neurons behave, for example when an animal is hungry or sleepy.
  • Some hidden connections: some kinds of connections, such as electrical synapses, are hard to see in these images.
  • Learning and change: connections change with experience; one map is a snapshot of one animal at one time.
  • Differences between individuals: one animal's map is not exactly the same as another's.
Python: same wiring, different behaviour (made-up)python
import numpy as np

# MADE-UP: identical wiring, but a "hunger" signal scales one connection.
W = np.array([[0.0, 0.8],    # smell -> approach
              [0.0, 0.0]])
smell = np.array([1.0, 0.0])
for hunger_gain in (1.0, 1.5):
    drive = (smell @ (W * [[1, hunger_gain], [1, 1]]))[1]
    print(f"hunger gain {hunger_gain}: approach drive {drive:.2f} ->",
          "APPROACH" if drive >= 1.0 else "ignore")
Note
This is why scientists combine maps with recordings of activity, experiments and models. The wiring is the foundation, not the whole building.

Ethics of brain data

Fly brain maps raise few privacy worries. Human brain data is different. Recordings from brain devices, and even brain scans, may one day reveal things a person has not chosen to share, such as health conditions or reactions. Some questions to ask:
  1. 1

    Consent

    Did the person clearly agree, and do they understand what the data may reveal later?
  2. 2

    Purpose

    Is the data used only for what was agreed, such as medical care or research?
  3. 3

    Security

    Is it stored safely, with access limited to the people who need it?
  4. 4

    Fairness

    Could it be used to judge people unfairly, for example in jobs, insurance or schools?
  5. 5

    Rights

    Can people see, correct or delete their data?
Some countries and international bodies are discussing special protections for brain data, sometimes called "neural data" or "neurorights". Rules differ by country and are changing, so check the current rules where you live. In India, general personal data law applies; check current guidance on how it treats health and brain data.
Watch out
Be wary of products that claim to "read your mind" or measure your intelligence or honesty from brain signals. Ask what the device actually measures, what evidence supports the claim, and where your data goes. For any medical question, talk to a qualified doctor.
Do this
Open science made the fly map possible: shared images, shared tools and many volunteers. Keeping human brain research open and protecting people's privacy is the balance the field is working towards.

Watch and read more

Lab

A short experiment showing that the same made-up wiring can behave differently, and a one-page note on the limits of wiring maps and the ethics of brain data.

~30 min

The problem

This runs in your browser. The network is made up.

  1. Run the starter. The same three-neuron wiring decides between "eat" and "rest".
  2. Change only the modulation numbers (not the wiring) to find one setting where the fly eats and one where it rests. Write down both.
  3. Write a short note (8–12 lines): three things a wiring diagram alone does not tell you, with one example each from this track.
  4. Add 4–6 lines: if a company wanted to collect brain signals from students to "measure attention" in class, what questions would you ask about consent, purpose, security and fairness?

Practice lab

Edit the code and run it. Python runs in your browser — nothing is sent to our server.

Output

No output. Add print(...) to see values.

You are done when

Hard questions

Try to answer before you reveal. If you can answer these, you understood the lesson.

Q1The FlyWire-based model in lesson 3 made useful predictions from wiring. Does that contradict this lesson's warning that wiring is not enough?Reveal

No. Both are true. The model shows that wiring carries a lot of information, enough to predict some behaviours that experiments confirmed. But it also needed assumptions about neuron behaviour and connection signs, and it could not capture states like hunger unless they were added. Wiring is a strong foundation; other measurements fill in what it leaves out.

Q2Why might brain data need stronger protection than, say, your shopping history?Reveal

Brain data may reveal things about health, emotions or reactions that a person has not chosen to share, and possibly things they are not even aware of. It is hard to change or "reset" like a password, and it could be misused to judge people in jobs, schools or insurance. That is why many experts argue for clear consent, strict purpose limits and strong security.

Please sign in to continue.

Questions people ask

Should AI copy the brain exactly?

Not necessarily. Like aeroplanes and birds, the most useful path is often to borrow the key idea and build it in a way that suits the technology. Some researchers do aim for very brain-like models, mainly to understand the brain itself.

If we had a human connectome, could we copy a person's mind?

Nobody knows how to do that, and this lesson shows why a wiring map alone would not be enough: activity, chemistry, learning and much more also matter. Be sceptical of claims that a person's mind can be uploaded.

Is my data from a fitness band or meditation headband "brain data"?

A fitness band usually measures things like heart rate and movement, not the brain. Some consumer headbands do record brain signals. Either way, read what the device records, where the data is stored, and who it is shared with, and check the current privacy settings.

How can I start in neuroscience and AI as a student?

Learn Python and basic maths, try simple neuron models like the one in lesson 5, and explore open datasets such as the FlyWire Codex. Many universities offer courses in computational neuroscience; check current programmes and free online courses.

Lesson test

5 questions. Get 3 right (60%) to pass and complete this lesson.

Sign in to take the test and save your progress