blockchains for artificial intelligence
TRANSCRIPT
![Page 1: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/1.jpg)
Blockchains for Artificial Intelligence
Trent McConaghy
@trentmc0
![Page 2: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/2.jpg)
Context:
Databases,Compute Stack,Decentralization
![Page 3: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/3.jpg)
What’s the difference between a database and a spreadsheet?Querying. From M’s of records, find the relevant ones.
1 Line of standard code, optimizedVs
50-500 lines of slow custom code, unoptimized
![Page 4: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/4.jpg)
The first “Blue Ocean” DBs: Relational DBsBenefits: powerful structured querying
Winner: Oracle, 80s and 90s
![Page 5: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/5.jpg)
The next “Blue Ocean” DB:Website-ready DBs
New benefits: lightweight for startupsWinner: MySQL, early 2000s
![Page 6: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/6.jpg)
The next “Blue Ocean” DB: Distributed / NoSQL DBsNew benefits: “Big data” scale, flexible schemasWinner: MongoDB, late 2000s-now
![Page 7: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/7.jpg)
Databases in status quo compute stack
FILE SYSTEMe.g. S3, HDFS
APPLICATION
PROCESSINGe.g. EC2
DATABASEe.g. MySQL, MongoDB
PLATFORMe.g. AWS, Azure
CO
NN
ECT
NET
WO
RK
S e.
g. T
CP
/IP
![Page 8: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/8.jpg)
FILE SYSTEMe.g. S3, HDFS
APPLICATION
Towards a decentralized compute stack
PROCESSINGe.g. EC2
DATABASEe.g. MySQL, MongoDB
Bitcoin blockchain?
PLATFORMe.g. AWS, Azure
CO
NN
ECT
NET
WO
RK
S e.
g. T
CP
/IP
![Page 9: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/9.jpg)
e-gold / e-cashBitcoinC
ON
NEC
T N
ETW
OR
KS
e.g.
TC
P/I
P
FILE SYSTEMe.g. S3, HDFS
APPLICATION
PLATFORMe.g. AWS, Azure
PROCESSINGe.g. EC2
DATABASEe.g. MySQL, MongoDB
Towards a decentralized compute stack
![Page 10: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/10.jpg)
e-gold / e-cashBitcoin, zcashC
ON
NEC
T N
ETW
OR
KS
e.g
. TC
P/I
P, Interledger
FILE SYSTEMe.g. S3, HDFS, IPFS
DATABASEe.g. MySQL, MongoDB
???
APPLICATION
PLATFORMe.g. AWS, Azure, Eris/Monax, BlockApps
PROCESSINGe.g. EC2, Ethereum, Hyperledger, Tendermint, Lisk
Towards a decentralized compute stack
![Page 11: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/11.jpg)
Benefits of blockchain technologyDecentralized / shared control
Immutability / audit trailTokens / exchanges
![Page 12: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/12.jpg)
Past blue ocean DBs: SQL, Internet, NoSQLThe next blue ocean DB: blockchain database
New benefits: decentralized, immutable, native assetsWho: BigchainDB
![Page 13: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/13.jpg)
How to build a scalable blockchain database (BigchainDB)
1. Start with an enterprise-grade distributed DB2. Engineer in blockchain characteristics
• Decentralized – each node has different sysadmin, ..• Immutable – Merkle chain of hashes, continuous backups, ..• Native assets – have private key = own, transactions, ..
![Page 14: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/14.jpg)
IPDB = a public global blockchain database
![Page 15: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/15.jpg)
e-gold / e-cashBitcoin, zcashC
ON
NEC
T N
ETW
OR
KS
e.g
. TC
P/I
P, Interledger
FILE SYSTEMe.g. S3, HDFS, IPFS
DATABASEe.g. MySQL, MongoDB
BigchainDB + IPDB
APPLICATION
PLATFORMe.g. AWS, Azure, Eris/Monax, BlockApps
PROCESSINGe.g. EC2, Ethereum, Hyperledger, Tendermint, Lisk
Towards a decentralized compute stack
![Page 16: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/16.jpg)
Introduction to AI,By Example
![Page 17: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/17.jpg)
[ibm.com]
What’s Artificial Intelligence (AI)?Original: AI: “A machine that can replicate human cognitive
behavior” [Turing test]
Now: AI: “A machine that can perform a cognitive task,
that was previously only possible with a human” [AlphaGo]
+ “A machine that can perform a non-analytical
information processing task, at speed / accuracy / capacity not possible by a human.” [Driving Moore’s Law]
![Page 18: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/18.jpg)
What’s Artificial Intelligence (AI)?
AI has a toolbox of ways to solve:• Classification• Regression• Density estimation• Rare event estimation• Knowledge extraction• Optimization • Active analytics• Structural design• Control / autonomy• …
For application to:• Operations• Product• Services
Example biz benefits:• Shorter time to market• Lower cost• Higher yield• New features possible• New products possible
Let’s explore these more…
![Page 19: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/19.jpg)
Classification, in 2DUse case: credit card applications
Salary
Age
Salary
Age
Didn’t pay
Paid bills
Credit profile:
How: linear classifiers, neural networks / deep learning, support vector machines, boosted trees, random forests / bagged trees
![Page 20: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/20.jpg)
Classification in 20DUse case: audio radar for tanks [McC ’97 Natl Def.]
![Page 21: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/21.jpg)
Classification in 10,000 DUse case: yield of memory columns [McC ’12 Solido]
![Page 22: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/22.jpg)
Regression, in 1D
age
job
satisfaction
job
satisfaction
![Page 23: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/23.jpg)
Regression, in 2D
![Page 24: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/24.jpg)
Regression in 10,000DUse case: reverse image search [Google]
![Page 25: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/25.jpg)
Regression in 10,000DUse case: reverse image search with provenance [McC ‘15 whereonthe.net // BigchainDB]
“Where’s Doge?”
![Page 26: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/26.jpg)
Density Estimation // Fitting a PDFUse case: 3-sigma circuit analysis [McC ’08 Solido]
![Page 27: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/27.jpg)
Density estimation // Fitting a PDFUse case: regression with Confidence Intervals, for predictive circuit design [McC ’11 Solido]
![Page 28: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/28.jpg)
Rare event estimation / “Black swan” simulationUse Case: 6-sigma analysis of bitcells [McC ‘09 Solido]
What is probability of this (rare) event
happening?o
utp
ut
![Page 29: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/29.jpg)
Y
X
Y
X
Knowledge extractionUse case: scientific modeling of analog ccts [McC ’05 KUL]
![Page 30: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/30.jpg)
1st split:
gain < 35 dB?2nd split:
GBW < 107 MHz?
Knowledge extractionUse Case: Topology decision tree [McC ‘08 KUL]
![Page 31: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/31.jpg)
Optimization
“Find the x that maximizes f(x)”(With as few evaluations of f(x) as possible)
x
f(x)
Find this
![Page 32: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/32.jpg)
OptimizationUse Case: Optimize perf. of lg. analog circuits [McC ‘01 ADA]
![Page 33: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/33.jpg)
OptimizationUse Case: verify circuits for worst-case perf. [McC ’09 Solido]
Maximum power
(Optimization runtime)
![Page 34: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/34.jpg)
Active Analytics“Find a mapping and visualize it, subject to high-d and expensive samples”
x1
x2
x1
2d 3d
x1
x1
f(x)
1d
x1x3
n-d
?
?
x1 x3
x2
x2
[stackexchange.com]
f(x)
x2
![Page 35: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/35.jpg)
Active AnalyticsUse Case: Interactive Circuit Design [McC ’12 Solido]
Decision variables are ranked by importance. User selects variable(s)…
… to plot predictions of outcomes at selected variables
Under the hood, a machine learning engine adaptively samples the space of possible decisions, and measures outcomes.
… and explore decision space (by dragging the orange crosshairs)
![Page 36: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/36.jpg)
More Applications of Optimization & Analytics
Business Intelligence: optimize churn & other key performance indicators (KPIs)
Big data infrastructure: optimize reliability / uptime, minimize power consumption, ..
Telecom infrastructure: capital & resource allocation
Internet / mobile: auto SEO, optimize for app store placement (rank, profitability)
ML modeling: Find optimal model meta-parameters (DeepNN, RF, SVM, ..), for application to computer vision etc.
![Page 37: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/37.jpg)
Structural Design // Machine CreativityUse Case: antenna design [Hornby & Lohn ‘04 NASA]
![Page 38: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/38.jpg)
Structural Design // Machine CreativityUse Case: quantum computing algorithm design [Spector ’04]
![Page 39: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/39.jpg)
Structural Design // Machine CreativityUse Case: circuit topology design [McC ‘06 KUL]
![Page 40: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/40.jpg)
Structural Design // Machine CreativityUse Case: jewelry design [Hornby ‘11 Orbimi]
![Page 41: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/41.jpg)
Example: A Single Product Suite With Broad Use of AI[McC Solido ‘05-13]
Synopsys HSPICE®/ CustomSim™
Cadence Spectre®/APS
Mentor Eldo®
Synopsys FineSim™
Agilent GoldenGate
Berkeley Design Analog FastSPICE™
TSMC
GLOBALFOUNDRIES
Samsung
Cadence Virtuoso® ADE
Netlist
High-Sigma MCFast, accurate,
scalable, verifiable
6σ Monte Carlo
Parallel Simulation InterfacePDK Interface Environment Interface
Cell OptimizerAuto variation-aware design space
exploration of memory/std cells
Fast PVT2-50X faster
verification across
PVT corners
Fast MC2-10x faster 3s
verification,
statistical corners
Fast Design SweepFast, thorough manual variation-aware
design space exploration
Solido Variation DesignerHier. MC
Fast statistical
memory array /
column analysis
• Regression with interpolation & CIs (KRC: scalability via divide-and-conquer on GPM)
• Model-based optimization, reliably finds global optimum by accounting for error in CIs
• Parallel computing
• 1-d density estimation (extrapolate via NQ)• Low-discrepancy sampling (High dimensionality via
modified Lattice Rules)• Data mining for variable sensitivities• Fast-evaluation opt. (evolutionary progr.)• Regression w/ interpolation; model-based opt.• Parallel computing
• Rare-event estimation (HSMC algorithm: transform into ranking problem, solve with adaptive sampling)
• High-dimensional regression (FFX: pathwise learning on huge # basis functions)
• High-dimensional classification (FFXC: pathwise ..)• Data mining for variable sensitivities• Parallel computing
• Model-based optimization• Regression with
interpolation & CIs (KRC: scalability via divide-and-conquer on GPM)
• Parallel computing
• Active learning via model-based optimization• Regression with interpolation & CIs (KRC:
scalability via divide-and-conquer on GPM)• High-dimensional visualization / sweep
exploration• Data mining for variable sensitivities• Data mining for variable-interaction sensitivities• Parallel computing
• MC sampling on hierarchically organized design (Fast Hier MC algorithm: transform into ranking problem, solve with adaptive sampling)
• High-dimensional regression (FFX) • High-dimensional classification (FFXC)• Data mining for variable sensitivities• Parallel computing
![Page 42: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/42.jpg)
Solves:• How to design with 1 billion devices?• How to design when the physics gets crazier at every process
generation?
AI is driving Moore’s Law[Via Solido McC] and others
12
14
16
18
20
22
24
26
28
30
2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
Year of production
Gat
e le
ng
th (
nm
)
![Page 43: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/43.jpg)
Recap so far
AI has a toolbox of ways to solve:• Classification• Regression• Density estimation• Rare event estimation• Knowledge extraction• Optimization • Active analytics• Structural design• Control / autonomy• …
For application to:• Operations• Product• Services
Example biz benefits:• Shorter time to market• Lower cost• Higher yield• New features possible• New products possible
![Page 44: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/44.jpg)
AI Sub-Fields• machine learning• neural networks / deep learning• evolutionary computation (GAs, GP, PSO, ..)• artificial general intelligence (AGI)• pitfall to avoid: getting caught up in the latest
buzzwords (it happens all the time.)
Strong relations to:• statistics (“ML is modern statistics”), probability• linear algebra (“flow of tensors” – TensorFlow)• nonlinear programming, optimization• control systems / cybernetics• Monte Carlo methods• philosophy, ethics (friendly AIs, AI rights, ..)
![Page 45: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/45.jpg)
How can blockchains help AI?
![Page 46: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/46.jpg)
Work off of each of the benefits…Decentralized / shared control
Immutability / audit trailTokens / exchanges
![Page 47: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/47.jpg)
Decentralized / shared control encourages data sharingMore data better models
-10.3 + 7.08e-5 / id1 + 1.87 * ln( -1.95e+9 + 1.00e+10
/ (vsg1*vsg3) + 1.42e+9 *(vds2*vsd5) / (vsg1*vgs2*vsg5*id2) )
10^( 5.68 - 0.03 * vsg1 / vds2 -55.43 * id1+ 5.63e-6 / id1 )
90.5 + 190.6 * id1 / vsg1 + 22.2 * id2 / vds2
2.36e+7 + 1.95e+4 * id2 / id1 - 104.69 / id2 + 2.15e+9 * id2 + 4.63e+8 * id1
- 5.72e+7 -2.50e+11 * (id1*id2) / vgs2 + 5.53e+6 * vds2 / vgs2 + 109.72 /
id1
Merge
Build models Build model
Low accuracyHigh accuracy
![Page 48: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/48.jpg)
Decentralized / shared control encourages data sharing Qualitatively new ecosystem-level data qualitatively new models
Example: shared diamond certification houses data makes fraud id possible
Merge
All the diamond cert houses
All
dia
mo
nd
s fo
r ce
rt h
ou
se 1
Ce
rt h
ou
se 2
Ce
rt h
ou
se 3
Ce
rt h
ou
se 4
All the legit diamonds
Build 1-class classifier
Legit
FraudulentNo single cert
house has enough data to make an
accurate classifier
Build classifiers
![Page 49: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/49.jpg)
Decentralized / shared control encourages data sharing Qualitatively new planet-level data qualitatively new models
“IPDB is kibbles for AI”
--David Holtzman
![Page 50: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/50.jpg)
Immutability for An Audit Trail on Training/Testing Data & ModelsFor greater trustworthiness of the data & models (Avoid garbage-in, garbage-out)
Provenance in building models:
• Sensor / input stream data
• Training X/y data
• Model building convergence
Provenance in testing / in the field:
• Testing X data
• Model simulation
• Testing yhat data Tim
e-s
tam
p /
sto
re
Applications:• you can tell if a sensor is lying• you know the “story” of a model• catch leaks in the data chain
![Page 51: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/51.jpg)
Another Opportunity:A shared global registry of training data & models
All the Kaggle datasets
All the Kaggle models
All the ImageNet datasets
All the ImageNet models
….…
.“Models are owned
by the planet”
![Page 52: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/52.jpg)
Training/testing data & models as intellectual property assets Decentralized data & model exchanges
Your datasets or models…
…licensed to others
Others’ datasets & models
…licensed to you
….…
.“EMX – European
Model Exchange?”
![Page 53: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/53.jpg)
Sell your CARTS?
![Page 54: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/54.jpg)
One more app AI DAOs
![Page 55: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/55.jpg)
Then you get decentralized processing.
aka “smart contracts”
What if you used a blockchain to store state of a state machine?
State
Virtual machine
![Page 56: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/56.jpg)
Then you get decentralized processing.
And you can build a
world computer
having decentralized processing, storage, and communications
(e.g. Ethereum vision)
What if you used a blockchain to store state of a state machine?
State
Virtual machine
Decentralized applications (dapps)
World computer
![Page 57: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/57.jpg)
DAO: a computational process that
• runs autonomously,
• on decentralized infrastructure,
• with resource manipulation.
It’s code that can own stuff!
DAO: Decentralized Autonomous Organization
State
Virtual machine
DAO Dapp
![Page 58: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/58.jpg)
AI entity is a feedback control system. That is, AGI.
Its feedback loop would continue on its own, taking inputs, updating its state, and actuating outputs, with the resources to do so continually.
AGI on a DAO?
AI DAO
World computer
![Page 59: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/59.jpg)
Example: The ArtDAOAlgorithm…
1. Run AI art engine to generate new image, using GP or deep
2. Claim attribution in blockchain, using ascribe
3. Create multiple editions, using ascribe
4. Post editions for sale onto a marketplace, using Getty (centralized), or OpenBazaar (decent.)
5. Sell the editions. $ goes to ArtDAO using built-in cryptocurrency like Ether. IP go from ArtDAO using ascribe.
6. Repeat! Create more art, sell it, get wealthier
![Page 60: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/60.jpg)
Example: The ArtDAOAlgorithm…
1. Run AI art engine to generate new image, using GP or deep
2. Claim attribution in blockchain, using ascribe
3. Create multiple editions, using ascribe
4. Post editions for sale onto a marketplace, using Getty (centralized), or OpenBazaar (decent.)
5. Sell the editions. $ goes to ArtDAO using built-in cryptocurrency like Ether. IP go from ArtDAO using ascribe.
6. Repeat! Create more art, sell it, get wealthier
Over time, if ArtDAO makes more money from sales
than from generating new art, then
it will accumulate wealth. And, you can’t turn it off.
![Page 61: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/61.jpg)
Angles to Making AI DAOs
• DAO AI DAO. Start with DAO, add AI. E.g. Plantoid
• AI AI DAO. Start with AI, add DAO. E.g. numer.ai
• SaaS DAO AI DAO. Convert SaaS to DAO. Then add AI
• Physical service AI DAO. E.g. Uber self-owning cars
![Page 62: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/62.jpg)
AI DAOs II
![Page 63: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/63.jpg)
Giving AIs control of our cars
![Page 64: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/64.jpg)
Giving AIs control of our trucks
![Page 65: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/65.jpg)
Giving AIs control of our grid
![Page 66: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/66.jpg)
All in the name of optimization
![Page 67: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/67.jpg)
Higher-level integration, therefore higher-level optimization
![Page 68: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/68.jpg)
AIs Become AI DAOs, everywhere
The AIs now own stuff.You can’t turn them off.
![Page 69: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/69.jpg)
AIs Become AI DAOs, everywhere
The AIs now own stuff.You can’t turn them off.
And you’ve just given them control of all your resources.
(Oops)
![Page 70: Blockchains for Artificial Intelligence](https://reader033.vdocuments.site/reader033/viewer/2022051201/58a1a3181a28ab88438b945e/html5/thumbnails/70.jpg)
Blockchains for Artificial Intelligence
A planetary-scale blockchain database (IPDB) unlocks opportunities:1. Data sharing Better models2. Data sharing Qualitatively new models3. Audit trails on data & models for more trustworthy predictions4. Shared global registry of training data & models5. Data & models as IP assets data & model exchange6. AI DAOs – AI that can accumulate wealth, that you can’t turn off
(be careful)
Trent McConaghy@trentmc0
bigchaindb.comipdb.foundation