Code Interpreter.... but OPEN SOURCE? Open Interpreter's Mike Bird on OS Projects, Mindset + More

Summary

In this engaging conversation, Mike Bird discusses the potential and challenges of Open Interpreter, an open-source alternative to OpenAI's code interpreter. He shares his journey from discovering Open Interpreter to becoming a part of its team, highlighting the importance of open-source contributions in AI. Mike emphasizes the need for a mindset shift towards exploration and experimentation when working with AI tools. He also touches on the impact of AI in various industries, the balance between innovation and safety, and the future of AI technology in making tasks more accessible and improving quality of life.

Takeaways

• 🚀 Open Interpreter is an open-source version of OpenAI's code interpreter, aiming to bridge the gap between natural language and computer control.
• 🌐 The project was launched in September and has since attracted a technical audience interested in leveraging its capabilities.
• 📱 Mike Bird was the first person to successfully run Open Interpreter on a phone, which gained attention and led to further involvement with the project.
• 🔍 Open Interpreter is still in its early stages (version 0.2), and the team is focused on developing the product to fulfill its potential.
• 🛠️ The tool is currently available via a command-line interface or as a Python library, catering to a tech-savvy audience.
• 🔧 Open Interpreter's auto-run feature is disabled by default, ensuring a human is always in the loop to monitor and guide the AI's actions.
• 🤖 The potential applications of Open Interpreter are vast, from coding assistance to automating mundane tasks, improving productivity and quality of life.
• 🌟 Mike Bird's passion for open source stems from the belief in sharing creativity and enabling others to build upon shared knowledge without starting from scratch.
• 🌐 The open-source community, particularly through platforms like Discord and GitHub, plays a crucial role in supporting and advancing projects like Open Interpreter.
• 🔄 Open Interpreter's approach to AI is about removing barriers and making technology more accessible and inclusive for everyone.
• 🌱 The future of Open Interpreter includes the development of a desktop app and the possibility of running the tool completely locally on various hardware devices.

Q & A

What is the main purpose of Open Interpreter?

Open Interpreter is an open-source tool that allows Local AI to run on your computer, serving as an interface between natural language and computer control to accomplish various tasks.

How did Mike Bird first get involved with Open Interpreter?

Mike Bird discovered Open Interpreter shortly after its launch and started experimenting with it. He became the first person to run it successfully on a phone, which gained some attention. Eventually, the creator of Open Interpreter, Killian, invited Mike to join the team.

What is the current state of Open Interpreter in terms of development?

Open Interpreter is still in its early development stages, with the current version being 0.2. It is primarily targeted at a technical audience and is available via a command-line interface tool or as a Python library.

What are some of the challenges faced by Open Interpreter?

One of the main challenges is that it is very technical and intimidating for most people due to its command-line interface. Additionally, the tool is still in its infancy and needs further development to fulfill its potential.

How does Mike Bird use AI in his daily life?

Mike Bird uses AI primarily for coding and offloading small, tedious tasks. He also uses it to learn new tools and for automating processes, such as setting up his work environment with a Stream Deck.

What is the importance of mindset shift for people using open-source tools?

The mindset shift involves a willingness to explore, investigate, and experiment with new tools and models. Users should understand that these tools are not fully polished and that best practices are still being defined.

How does Mike Bird stay positive about AI and its potential despite the negative headlines?

Mike focuses on the positive aspects and potential benefits of AI, such as advancements in medical technology and improvements in quality of life. He believes that the positive outcomes outweigh the negative and that society should embrace the technology responsibly.

What are some resources Mike Bird recommends for staying updated on AI news?

Mike recommends following selected accounts on Twitter, subscribing to newsletters like Ben's Bites, and joining Discord communities like the Open Interpreter Discord to stay updated on AI news and trends.

What is the significance of the open-source approach in the development of AI tools like Open Interpreter?

The open-source approach allows for collaboration and rapid innovation. It enables developers to build upon each other's work, share knowledge, and create tools that can benefit a wider audience without being restricted by proprietary barriers.

What are the future plans for Open Interpreter?

There are plans to release an open-source equivalent of the Rabbit R1, a hardware device that would allow Open Interpreter to run on a physical device, making it easier for users to interact with and utilize the tool.

Outlines

00:00

🚀 Open Interpreter and Local AI

The conversation starts with Mike Bird discussing the potential of distilling vast information into trends, specifically mentioning Discord and open-source projects. Mike shares his journey with Open Interpreter, an open-source version of OpenAI's code interpreter, which allows running code on a computer to accomplish various tasks. He highlights the importance of natural language control over computers and the future potential of AI in making computing more accessible and inclusive.

05:01

🧠 Mindset Shift for Open Source Tools

The discussion shifts to the mindset required for effectively using open-source tools. Mike emphasizes the need for exploration, investigation, and experimentation. He talks about the technical audience for Open Interpreter and the challenges of using command line interfaces and Python libraries. Mike also mentions the development of a desktop app and the importance of managing expectations when working with tools still in their infancy.

10:01

🌐 Open Source Models and Real-World Use Cases

Mike delves into the best use cases for different open-source models, discussing the importance of understanding the specific capabilities of each model. He mentions tools like Jan and Olama for switching between models and the concept of an 'arena' for ranking models based on user preferences. Mike also touches on the commercialization of open-source projects and the benefits of releasing work as open source.

15:03

🌟 Positivity and AI Innovation

The conversation focuses on maintaining positivity in the face of AI innovation. Mike shares his enthusiasm for the potential benefits of AI, such as advancements in medical technology. He discusses the balance between progress and caution, emphasizing the need for both 'doomers' and 'risk-takers' in society. Mike also talks about his personal use of AI for coding and automation, and how it has improved his productivity and quality of life.

20:05

🔍 Navigating AI News and Information

Mike shares his approach to staying updated with AI news, highlighting the importance of curating sources like Twitter and newsletters. He mentions specific resources like Ben's Bites and AI Breakdown, as well as the value of Discord communities for staying connected with the latest developments. Mike also discusses the challenge of managing information overload and the need for focus in the rapidly evolving AI field.

Keywords

💡Open Source

Open Source refers to something that can be freely used, modified, and shared by anyone. In the context of the video, it relates to the philosophy and practice of allowing others to view, use, and improve upon the source code of software. The guest, Mike Bird, is passionate about open source projects like Open Interpreter, which he sees as a way to democratize access to technology and accelerate innovation. The video discusses the benefits of open source, such as community collaboration and the ability to build upon the work of others without starting from scratch.

💡Open Interpreter

Open Interpreter is an open-source project that serves as an interface between natural language and computer control, allowing users to run code on their computers to accomplish tasks. It is mentioned in the video as a tool that has the potential to impact various industries by offloading mundane work and enabling more efficient and automated processes. Mike Bird, the guest, has joined the Open Interpreter team and is excited about its future, emphasizing its role in making AI more accessible and inclusive.

💡AI (Artificial Intelligence)

Artificial Intelligence, or AI, refers to the simulation of human intelligence in machines that are programmed to think and learn like humans. In the video, AI is discussed as a transformative technology with the potential to improve various aspects of life, from medical advancements to productivity tools. The guest talks about his work with AI, particularly in the context of local AI and open-source models, and how they can be used to solve complex problems and automate tasks.

💡Local AI

Local AI refers to AI systems that operate on local devices, such as personal computers or smartphones, rather than relying on cloud-based services. The video discusses the benefits of local AI, including privacy, reduced dependence on internet connectivity, and the ability to run AI models offline. Mike Bird shares his journey into local AI and the challenges of making these systems more accessible and user-friendly.

💡Natural Language Processing (NLP)

Natural Language Processing is a subfield of AI that deals with the interaction between computers and humans through natural language. In the video, NLP is crucial to the functionality of Open Interpreter, as it allows the system to understand and execute tasks based on natural language inputs. The guest explains how NLP enables the translation of human language into computer commands, making AI more accessible and intuitive to use.

💡Hackathon

A hackathon is an event where people, often programmers, collaborate intensively on a project. In the video, hackathons are mentioned as a way to foster innovation and community engagement. The guest talks about participating in hackathons related to AI and blockchain, and how these events can lead to the development of new ideas and solutions, such as the open-source version of the robot R1.

💡Community

In the context of the video, community refers to the group of people who contribute to and support open-source projects like Open Interpreter. The guest emphasizes the importance of community in driving the development of these projects, sharing knowledge, and providing support. The Open Interpreter community, for example, offers tutorials, guides, and a platform for discussion, which helps users learn and contribute to the project.

💡Innovation

Innovation is the process of creating new ideas, methods, or products. The video discusses how open-source projects and AI can drive innovation by providing tools and platforms for people to explore, experiment, and solve problems. The guest shares his experiences with hackathons and open-source projects, highlighting the role of collaboration and the willingness to explore in fostering innovation.

💡Automation

Automation refers to the use of technology to perform tasks automatically. In the video, the guest talks about using AI for automation, particularly in coding and offloading tedious tasks. This allows for increased productivity and efficiency, as well as freeing up time for more creative or leisure activities. The discussion touches on how AI can change the baseline of capabilities and remove mediocrity by enabling everyone to be more capable.

💡Research Time

Research time is the period dedicated to investigating, studying, and developing new ideas or solutions. The video emphasizes the value of research time in open-source projects, as it allows for deeper understanding and the development of more effective tools and technologies. The guest mentions how open-source projects prioritize research time, which can lead to significant advancements and improvements in various fields, including AI and medical technology.

Highlights

Open Interpreter is an open-source version of OpenAI's code interpreter, allowing LLMs to run code on your computer for various tasks.

The tool acts as an interface between natural language and computer control, offering a broad range of capabilities.

Open Interpreter was launched in September and has been seen as a significant step towards natural language control of computers.

The first person to successfully run Open Interpreter on a phone, which led to a viral tweet and further engagement with the community.

Open Interpreter is still in its early stages, with the current version being 0.2, and is primarily targeted at a technical audience.

The importance of mindset shift in adopting open-source tools, including a willingness to explore, investigate, and experiment.

The potential of using Google's Colab to leverage external GPUs for running resource-intensive AI models.

The concept of using different AI models for specific tasks, acknowledging that each model has its strengths and weaknesses.

The challenge of finding the best model for a given task, which is a dynamic process as new models are released regularly.

The idea of using AI to offload mundane tasks, allowing professionals like doctors to focus on research and innovation.

The impact of AI on improving quality of life and reducing suffering, as seen with advancements like AlphaFold in medical research.

The importance of balancing progress with responsible deployment, ensuring that AI technologies are used ethically and safely.

The potential of embedded AI to democratize access to AI technology, making it more accessible for everyday tasks.

The value of open-source projects in fostering innovation and collaboration, as well as the potential for commercial opportunities.

The role of open-source communities in providing support, tutorials, and a collaborative environment for users and developers.

The importance of staying up-to-date with AI news and developments, using platforms like Twitter, Discord, and newsletters.

The idea of distilling vast amounts of information into trends and summarizing them for easier consumption and further exploration.

Transcripts

that's where I see a lot of value um for

any entrepreneurial people listening

where if we can distill the vast amount

of information into what's trending in

Discord trending on X and be able to

like package it in a way where you can

link to more information that's a very

valuable product and I hope it's open

sourced Renee here at UNS supervised

learning your easy listening podcast for

bleeding edge open source Tech I'm

sitting with Mike Bird who I came across

on a I think it was a Twitter post that

was again calling out Jan and I got

notified it was like hey we should reach

out to this person and you're doing a

lot of really cool stuff with local Ai

and now you've joined open interpreter

officially um yeah so open interpreter

my understanding is it's an open version

of open ai's code interpreter which it

sounds like I've just said the word open

like 50 times but do you want to take it

away yeah we're the legit open one so

open interpreter is a way for llms to be

able to run code on your computer to

accomplish tasks it you can think of as

like the interface between natural

language and computer control and what

do you use it for so one cool thing with

open interpreter is it's just the

breadth of capability uh it's really

like the everything app we've just had

to you know make the pilgrimage but you

were saying about how you came to be at

open

interpreter yeah so um open interpreter

was launched last September uh early on

and I immediately saw and thought like

this is the future the ability to take

natural language and control computer is

something that everyone's kind of

working towards and this was that first

like big leap towards that so right off

the bat just started playing with it

open source anyone can download it try

it and I just started playing around and

really enjoyed like the capabilities

that unlocked I was the first person to

get it running successfully on a phone

and that demo actually caught a little

bit of fire it was my first viral tweet

really really excited about it uh and it

was super basic but then Killian the the

creator of open interpreter uh reached

out and said hey the next house party we

have we have these digital House Parties

which are a lot of fun uh he said you

want to give another demo so I said

absolutely of course put together a

little bit more uh gave a demo there

everyone really liked it and then just

kind of kept chipping away um I've been

doing freelance work for the last year

and a bit um and then just doing a on

the side I've either been a software

developer or the AI guy at companies and

then staying up to date on what the open

source world was doing what tools were

available just eventually making some

demos trying out different things it

just kept getting more and more Steam

and then one day kilan was like hey do

you want to uh join the team like

absolutely I've been doing software

since 2015 and always had the passion

for open source it's it's a gift to to

the world being able to just like take

your creativity put it out there and

then ever the developer gets to start

from from that point and move ahead

rather than having to start from scratch

so the ability to work for an open

source project especially when as cool

as open interpreter has just been

absolutely awesome so yeah it's just one

of those passion projects where uh an a

good idea that everyone kind of uses

like the natural progression of how we

interface with computers remove barriers

allow more accessibility inclusivity in

access to compute and open interpreters

making sure everyone has access to that

and it's not hidden behind you know an

open AI pay wall have you found and so

my journey into getting into local AI

was a little bit of it was through like

no code and wanting to do things without

open ai's API and realizing that you can

actually run things offline so what what

are you finding most people's use cases

is it a fairly Tech heavy audience yeah

yeah definitely in uh at this point it's

very Tech heavy we only offer or open

interpreter is only available via uh a

command line interface tool or as a

python library and opening up that

little black box with just a flashing

cursor is pretty intimidating for most

people uh there is a desktop app in the

work so you can join the weight list for

that but right now we are just making

sure that the product gets developed

we're still at version 0.2 still very

early days so the technical audience has

more understanding nature around how can

we work with this tool that's still in

its infancy to build it up into

something that has all the potential

that open interpreter promises and even

at this current stage the current state

of Open Source LMS are not quite at the

point where a wide range of capabilities

are there but the people who use the

open source models realize that like

each one is tailored to a specific thing

like you can use something like dolphin

mixol which is a very capable model

probably on par with GPD 3.5 but it's

still not going to reach the generality

of gp4 but then you get ones like SQL

coder which actually exceed GPT 4 in

turning natural language into SQL

queries so knowing that like having this

little vertical column of capability per

each model and being able to switch

between them with tools like Jan or

olama or llama file really allows a lot

of capability but we're not at the

general state that the general public

wants with a model like gbd4 however it

gets better every day so very soon um

we'll be able to run open interpreter

completely locally to the point where

it's able to do the tasks that you ask

of it on the whole like different models

for different tasks that's one of the

the

foundational kind of questions that I

see people asking so I had I don't know

if I wrote it in the questions that I

had prepared for you but it was kind of

what do you think is the key kind of

unlocks that people need to jump

from understanding so I had a a

recording yesterday where the person was

saying that GPT 4 helped them kind of

Leap Frog over in terms of their

technical capability where they wouldn't

have been able to do it without it like

their coding knowledge right and they're

not a coder but it's like are there are

there things that you think in terms of

like a mindset shift that people need to

make to be able to start working

specifically with open source tools I

would say the mindset shift needs to be

around a willingness to uh explore

investigate experiment play like just be

willing to approach a new tool a new

model with the understanding that it's

not going to have all the edges polished

everything is so new everything is

expanding so quickly that there's no

really defined way of best practices yet

so the understanding that using Model A

for something might get good results

model B might get great results but then

horrible results for something else and

it's not the fault of the model it's

just a very resource intensive process

to train fine-tune and deploy these

models even running them like the vast

majority of people's Home Hardware is

not at the same level as open AI server

Farms so making sure that expectations

are are leveled towards what we're

actually putting towards it because when

people talk about having a an an old

laptop with you know 16 gigs of RAM like

you're not going to be able to run that

much off of it and just measuring your

um expectations based on that but as you

can use tools like Google's collab or

runpod to actually leverage external

gpus then you start seeing like a big

increase in capability so it's kind of

the the flexibility with the existing

architecture of the open source

ecosystem versus what you want it to

accomplish

so if we if we go right back like I did

in a different podcast where I was like

okay let's go right back so Google

collab that enables end users to use

external gpus and that's the whole

renting gpus right yeah so with Google

collab it's an interface to write uh

jupyter notebooks python code and then

run it on Google um gpus so locally you

could have a MacBook with decent power

but through collab you can pay by the I

Believe by the second but let's say by

the minute or hour to get highend gpus

that you can really Crush through tasks

with and then you're just paying a bill

for a pay per use rather than having to

buy a little like tiny box or something

spend 15 grand to get it there you're

able to just pay per use and it does

enable you to access these things but

then you're still paying a a fair sum

and start using it regularly and that's

not locally run that's is that mixing

the two like it it's all yeah it's run

remotely so it's on the hosted by Google

but you can still use open source models

through that okay I'm going to have some

kind of like little diagram there for my

own understanding the other thing that

you said about different models I think

one of the that's one of the main

questions is like what's the best model

for this and it's like it's one of those

classic like it depends things so and it

changes every day because new models are

released every day but in my mind and I

think of it like a little very infantile

it's like this one's good for writing

this one's good for for coding so I

don't touch things like what is it deep

seek and those coding ones because I'm

like I have no use for it but do you

have uh like resources that you would

typically go to like a like a place

where you would recommend this is

generally a question that I leave to the

end but I just I'm really curious yeah

there's one uh I actually forget the

name I'll I'll grab it for you after but

there's an arena where what you'll do is

you go to this website and it will allow

you to ENT a prompt and then it'll

generate two responses and you just pick

which one's better if they're the same

you're mark them the same and in the

back end what they're doing is they're

ranking all the models based upon user

preferences and giving an ELO score

similar to how chess players are reigned

so with this it's really using the

general or like the actual output rather

than benchmarks because benchmarks can

be gained once you look at hugging faces

um like rankings based on benchmarks you

get these really small models that are

fine-tuned on exactly the test that

they're measured on and of course

they're going to crush it but as soon as

you take it out of that constrained

environment they're horrible so with

this Arena apprach approach you're

actually having users come in say I like

this output better than that output

you'll get models like uh Claude 2.1

ranking significantly lower than it's

claimed to just because it's so heavily

censored that its outputs are kind of

trash so getting real world use case

like that is good but one thing that's

very popular to people is Just Vibe

check download it play with it tools

like like Jan and AMA allow you to

quickly test different models side by

side where you can say hey this is what

I wanted to do I'm I'm a writer I needed

to track outlines for me I want a tool

that's really good for that and then you

just got to play with yourself it's one

of those things where you can't always

rely on the benchmarks because they can

be gamed pretty easily that's such a

common uh a common point that I'm

hearing across like doesn't matter who I

talk to really and so it's kind of

similar to like if I'm thinking about

reviews in like this because I'm coming

from the SAS world and it's like nobody

really trusts reviews anymore like G2 or

whatever it's like but when you

mentioned about the the patience it

makes me think of like commercialization

of Open Source and is open interpreter

that's not your first open source

project that you've been a part of it's

the first one that I've been employed

with I have worked at other companies

where we've released either um code

bases or full projects as open source um

I've been a big proponent I used to work

at a robotic software company and we had

a partnership with this robot

manufacturer and we said yep we we'll

gladly do this work for you but we want

to open source it and I super proud of

the company to do that because again

it's just it's very Niche not many

people will use it but now anyone can so

it just enables people to level up that

much quicker yeah because my my kind of

question it's not really question it's

not really a point it's just one of my

typical goes nowhere sentences with you

made me think about I I have to again

put it in the in the notes it was a

research paper on Game Theory of how why

people are so reluctant especially like

Enterprises are so reluctant to go open

source because they're like oh people

are going to steal my stuff and it's

like yeah well at what point it was like

infrastructure as a open source or in

I'm G to have to go back and edit this

out but it it was like have you

seen what kind of mindset shift

businesses have to make in order to be

kind of Pro open source or do you think

that it's not something that you can

repeat as a

pattern it's

definitely a different business model um

and a different mindset in general when

you look at companies like actually one

step back so the internet whether it's

servers or iot devices all run on Linux

which is an open source software and the

creative Linux is very doing very well

for himself because people understand

that there are add-ons you can do like

it's not just the software that making

money it's the services behind it and

that's kind of been my Approach too

where I am happy releasing the work that

I do as open source because people see

value in it people understand the

expertise that went into building it and

then they're like hey we would like you

to integrate this with our service hey

we would like to hire you to contract so

instead of just saying pay me for my

product it's here is an example here is

proof that I know what I'm doing now you

can bring me aboard and it's not for

everyone not every situation but the

people who can pull it off tend to have

uh a greater desire to kind of build up

the ecosystem as a whole and then the

ecosystem tends to them for it yeah

that's what I'm thinking because I I

just from like a brief interaction with

you you seem like a very positive person

which is I'm not saying that's unique in

open source but or in Tech but I see a

lot of um uh not not jaded people but

it's kind of how do you how do you

maintain that that positivity around the

innovation in AI right because there's a

lot of like doomers and then there's

also like how how are you remaining

balanced about

it it's very helpful being so ingrained

in it when you on the peripheral of Open

Source AI or just AI in general all

you're hearing is those Doom and Gloom

headlines being like the like end of

humanity all of these scary things but

when you're working on it directly you

you see the benefit like uh when Google

deep mine came out with Alpha fold all

of a sudden they solve the protein

folding problem which will lead to

countless medical advances which will

lead to countless saved but that makes

headlines for a day there's not people

arguing to like push open source or push

AI farther ahead because of these

advances like Alpha fold but the amount

of suffering that will be reduced the

amount of increasing quality of life all

over the world that AI will give is is

astronomical like there will be orders

of magnitude more Improvement in so many

people's lives because of AI getting

ingrained in more

services however that doesn't sell

headlines so a lot of people will just

click on to the the the anger and the

fear the manipulation that comes with

that and it's really unfortunate because

we're doing ourselves a disservice by

not sufficiently embracing a technology

and don't get me wrong we need some

degree of safests rather than doomers

who are dragging their feet a little bit

making sure we do things properly

because just like Society we need a push

and pull like we can't be 100%

Progressive or 100% conservative because

that is a recipe for failure we need a

little bit of both so having people say

look like before we just start deploying

this thing everywhere like let's make

sure we have like our BAS is covered

first very reasonable very responsible

we also need the people saying like we

need to take a little bit of risks like

there there's you got to risk it to get

the biscuit and it's very important as a

society that we understand that the the

improvements the the societal gains that

come from this are going to be so

beneficial that um so one example my uh

like my parents both retirement age I

live in a very small community with a

lot of like old people and every person

who say in 3 years dies of a disease

that AI cures in four years is a tragedy

we we should push it as fast as possible

in the medical realm to make sure that

we get these advances because the amount

of people who have been saved by

Advanced Medical Technology increases

every year and this can be a greatly

forward we'll have gene therapy will

have a whole bunch of mRNA um advances

that will really cure diseases that

right now don't have enough impact to

really justify the funding but AI is

going to all of a sudden change the

equation so

yeah staying positive is just kind of

keeping ey on the prize and realizing

that no matter what people say about the

negative aspects when you look at the

positive aspects it's it really

outweighs it I recently did like an

Asing interview with someone who worked

on a project to dis like it was to

discover the Parkinson's Gene via voice

in like low resourcing um areas over

over phone and it was it was like how

using AI I'm like how how do you do that

like I'm like the most non-technical

person I'm like that's incredible but I

see what you're saying like in terms of

your own because I'm I can't stop

looking at the 3D printer now that I

know what it is like your own AI

specific projects like what do you use

AI

for I use

it a lot of it's with coding a lot of it

is offloading the small tedious things

it also gets around trying to

develop automations I have here a little

a stream deck which is a cool device

that streamers use but I found It's a

Wonderful productivity tool and I had no

idea how to work it at first but AI can

kind of walk me through the process of

like understanding the tool learning the

tool it's a little like assistant like

you get the back and forth you get to do

the um Socratic method to learn things

and then when you understand you're like

okay I want to modify to suit my

expectations and that's where the coding

help comes in so it's still mostly in

the in the coding in the technical realm

um but then you can even Branch out

being like I have a buddy who's a

videographer and he does amazing work

just came out with a documentary which

is really cool to see but now ai enables

him to no longer rely on like external

editors external graphic designers AI

one of the cool things about it is it's

going to completely remove mediocrity

where the Baseline is now elevated

because AI enables everyone to be a

little bit more capable and that's one

thing that really excites me about open

interpreter because it has the ability

to impact every single industry

because you're able to offload work to

this little assistant here have it

create things for you have it accomplish

tasks for you you just give it natural

language I've been working on some ways

to automate that when you can just feed

it a spreadsheet as an example or any

document and then it parses through it

and accomplishes all of those tasks so

all of a sudden you're freed up to do

other things whether it's work more or

take more Leisure and then if you do do

the Leisure route all of a sudden like

your quality of life improves you become

a happier person less screen time

there's one thing to be said about how

much our current Society is stuck on on

screens and the internet and a lot of

that's because everything we're doing is

very manual we're not able to just say

find me a summary of all the news for

the day um you got to like look through

read it but if you have this little bot

that comes up and says these are the top

five points you should know you can

double click on something if you want

otherwise you can walk away it's going

to really free up time mental capacity

and allow us to you know get a more

balanced approach to life so my

understanding of open and has just

changed a little bit because now I'm

thinking of it like a almost like an AI

agent that has a is it called a

hierarchical understanding where you

pass it to it it'll do an activity and

it'll say is this okay and you'll be

like yeah otherwise it'll just go ahead

my yeah so yeah so one of the safety

mechanisms with open interpreter is auto

run is disabled by default so what what

open interpreter does is it'll take a

task develop a plan to to reach that

task or sorry reach that goal and then

write and execute the code to accomplish

it um but there's always a human in the

loop to make sure things are good you

can run it auto run uh like a

self-driving car keep your eyes on the

road like Don't Let It Go muck because

it might get into a loop drive up your

open AI Bill if you're using local model

make sure the alignment's good and then

otherwise it's it's a very capable tool

what's

alignment alignment is defined

differently by different people but you

can think of it as does the guard rails

on the AI align with your guard rails is

is the ai's goal aligned with your goal

so more often than not they don't have

goals unless you give them one

explicitly and then as long as that's

the case there's no issue so with tool

like open interpreter it will explicitly

lay out the plan and you can say hey

that is exactly what I want you to do

all good if not you just hit no and then

rephrase your request and then it can

adjust accordingly right I'm thinking

back to you said the stream deck that's

the thing with the little knobs right

yeah knobs and buttons so what what are

you using that for in everything from

like I got some lights here that I

automate I have a little I my work

button so as soon as I start my work day

I just hit it and everything pops up I

get logged in you can set different

automations I have my demo time so when

I want I press the button lights turn on

to to the right temperature got to get

the good skin color and then it opens up

my screen recording software it opens up

my code it opens up my terminal so it

just gets everything ready where instead

of having to do it's it's a few seconds

of time that it saves but it's

consistent results every time so I don't

have to worry about oh like did I tweak

this setting or did I forget to you know

open up this profile it's just the same

result every time so it's just one L

thing I have to worry about and I can

focus on the demo itself I wonder about

it that is really cool it's also I

literally recorded about home automation

yesterday um because it's one of my

goals as I've only just gotten Wi-Fi at

my house I've been hotspotting off my

phone for five years wow I know uh so

I'm very excited about lights but I I

have a tendency in automation to do

things that it will save me 15 seconds

of time and it takes me 15 hours to set

up the automation but I that's the

engineer dilemma we love that

stuff um like with open

interpreter you said that it's you can

use it through vs code or so uh use it

through the terminals so

it you almost think of it like a

standalone program so it can give you

code that you could copy and paste into

vs code but there are more optimized

tools whether it's so I use cursor as my

IDE and within cursor I use um Cody as

my inline text complete so those two AI

tools are very well optimized for uh an

editor for an IDE but open interpreters

more optimized for the general purpose

assistant okay and it's it's you

utilizing models that are like they open

models that any models that you want and

it's using natural language

understanding so it goes back to that

thing of like you can't you only kind of

get out what you put in so if I can't

succinctly explain what I want it's not

going to be able to do what I want like

correct and and there's also

configuration capabilities where you can

give it custom instructions so for

example I will say make sure you use ARC

browser or make sure you use as my

terminal instead of the default terminal

guiding it a little bit really helps but

even though special instructions the

importance of them vary between models

so again GPT 4 being state-ofthe-art is

a lot more capable at just kind of

inferring what you're looking for with a

little B of help some of the excuse me

open source models will require a little

more coaxing to do what you want so back

to what I said earlier like you need to

be willing to experiment and play with

the open source models because they

don't have the same

generality

yeah I wanted to ask like what sparked

you into exploring open

source it's yeah um it's always felt

right you know like the ability to code

is kind of like a superpower because you

take nothing a blank screen using your

brain you you create something that

people find value with like there

there's nothing there except like

changing ones and zeros you're

controlling the flow of electricity

through sand it it sounds like a

superpow five I've ever heard one and

then it's like to use your superpower

for are evil and not saying close source

is evil but it's constrainted in its its

reach and there tends to be a heavy

profit motive and I mean we need profit

motive to to encourage people to

dedicate their lives to things they

don't really want to do but like I said

I I feel I hit the jackpot getting this

job with open interpreter where I get to

spend my time and energy with my bills

covered being able to work towards

bringing this technology to everyone and

very fortunate not everyone's able to be

in a situation like that but when you

spend your life doing something that you

hate just to get a paycheck the Clos

vers open debate is less relevant but if

you're fortunate enough to be in a

position where it's like hey I get to

make the choices as to how I spend my

time um it just feels like the right

thing to do like there there are going

to be billions more people accessing the

internet and a portion of them are going

to be software developers and a portion

of those are going to be working with AI

and if we can enable them to have as

much tooling as possible they just like

ramp up and can contribute to This

Global e system of available software

that benefits

everybody I love your energy like

sincerely you mentioned about like not

being behind the screen do you do you

have outside Hobbies like I for bit of

context I live in a village of 600

people on uh peninsula in Ontario uh

about 3 hours north of Toronto um best

hiking best kaying you can find we're

we're Instagram famous for Crystal Blue

Waters um yeah spend the vast majority

of my time outdoors and then inside tend

to be my office yeah nice um who who

would you interview you mentioned the

Linux founder but who would you

interview in anyone in open source or Ai

and I have mentioned before that they

are allowed to be dead um um for AI I

would say I saw this someone else

mentioned this on your show so it's not

original but Andre karpathy he is he was

high up at Tesla hpid open AI he's been

involved with and overseen massive

advances in Ai and a wonderful

communicator educator if you're ever

looking to like understand llms more

deeply or want to build your own GPD his

YouTube channel is pure gold uh he also

seems like a cool guy to have a drink

with and you know I got to shoot my shot

to try to win over to team open source

so that would be cool and I actually

don't know if I'd want to interview Len

tald I feel like I'd walk away feeling

really dumb but I'll put a maybe pile

maybe pin on that one I I think um I

think that's one of the one of the best

things that I've done recently is like

the willingness to look dumb because you

can ask questions that people have had

on their mind and they're like oh I'm so

glad that she was the class clown for

that I thank you for your service that's

all right happy to do it uh your if you

could claim intellectual property rights

for anything or if you could uh I give

two options like you can claim

intellectual property rights or you can

have just invented anything what would

it be

me anything that's

so I kind of go back to my open source

stand I don't know so being in software

for this long I think software IP is

ridiculous the the fact that people can

try to claim things like rounded corners

and oneclick checkout is outrageous to

me like that's just a natural

progression of where software goes I

understand especially for like hardware

and products if you put your heart and

soul into something wanting to make sure

you reap the rewards of your creativity

in your hard work but for software no I

don't think there should be software

patents I think that's silly

um I don't know I so I have a a

background in in Health Sciences so for

me there's a big emphasis on medical

technology so if I could invent anything

it would definitely be something like

I'm drawing a blank on the name there's

the the nerdy Star Trek things where

like you can just like scan someone be

like oh like you have this going on like

this is how we fix you um the ability to

rapidly diagnose and treat

a multitude of diseases would be

something I would absolutely love to

participate in and I feel like

tangentially open interpreter is going

to be able to help reach those goals

because it'll be able to offload a lot

of the mundane work allowing doctors and

other um other people in the medical

profession to be more researchers rather

than having to like go through the

nitty-gritty like if we can have people

free up their time to be able to pursue

curiosity and investigate new potential

avenues for like treatment and diagnosis

uh is going to reduce suffering

immensely that's something that I've

realized in in open source like very

briefly being exposed to it is the the

value placed on Research time and um

yeah the value placed on Research time

and understanding and like being able to

sit with problems and also Gathering

different people's opinions like that is

so foreign to me because like I was I

was working with the Jan team and it was

so

um yeah so welcoming and just very like

I'm I'm not painting a picture of like

everything's perfect in open source but

it's I think it is that that thing that

you mentioned about like it being

beneficial to humankind like it's it's

better for the greater good like um but

with open interpreter I feel like I'm

getting this sense what are your

thoughts on like embedded AI is that the

right term even like because you kept

saying

robot yep so with the release of the

rabbit R1 uh Killian put out a call to

action being like let's make the open

source version to this and I don't know

when this episode's going out but

probably around that time there's going

to be a release of the 01 the open

source equivalent there's a group of

people meeting in Seattle every week

going through this six-week hackathon to

build this open source equivalent and

one of the purposes of that is to get

open interpreter on a hardware device

because if I could have a little box

right beside me I press the Buton and

say I need to do this task and I can

just offload it onto that little box and

then I do my own thing all of a sudden

I'm freed up to do so much more so I

don't think

it's uh like an end goal or like the

Grand Vision but it's a good component

in order to democratize access to this

technology that and that's in six weeks

time did you say Well they're pretty

deep into it so I don't want to put too

much pressure on them so let's say you

know around the end of the month

sometime in March maybe there'll be a

pretty big announcement a lot of people

have been involved with it it's uh it's

going to be pretty gamechanging and how

many hackathons have you like are you

into i' I've done a fair bit over my

over my career um first one was in

blockchain a couple other with AI um one

really cool thing that I've been

involved with is internal hackathons

where companies say you know what hands

off the keyboard like business processes

stop like let's do something that's

completely off the road map and just

like let your creativity flourish

and I firmly believe if more companies

Embrace this mentality a lot more

Innovation would be sparked because

going through the day and day out you

get a very deep understanding of the

product uh or service or whatever is

you're working with but there's always

that the human desire to kind of like

bring more or like explore little things

like back to the Curiosity so if more

companies were willing to just say look

let's get everyone together across

disciplinary team people with different

backgrounds different experiences and

say just make something cool like like

see what you can do and then and a lot

of really cool Innovations come from

that yeah I want to ask you one last

thing because again you seem so positive

and like you have actually slept and

most people in AI seem like they haven't

slept uh where are you like keeping up

with with AI news

and um honestly Twitter X is is by far

the best place in my mind there's you

you have to choose your accounts that

you follow wisely though because there

is just a fire hose of information um I

subscribed to a couple newsletters um if

you're looking for information on

tooling uh Ben's bites is pretty good um

if you're more into podcasts uh in

regards to like industry news I listen

to the AI breakdown da daily summary of

you know everything in the industry um a

bit less technical but still pretty good

summary um Discord has a lot of alpha

Discord if you can join the open

interpreter Discord and all of a sudden

you're just link to all these other

things um you just have to be able to

like parse through again the the fire

hose of information that's one of the

hardest things that's where I see a lot

of value um for any entrepreneurial

people listening where if we can distill

the vast amount of information into

what's trending in Discord trending on X

and be able to like package it in a way

where you can link to more information

that's a very valuable product and I

hope it's open sourc that's I that's the

the first thing that I've been trying to

do with AI agents and like experimenting

with Lang chain and I've seen uh I can't

remember her last name but Maya I

watched a video uh was essentially like

summarizing Discord Bots and like

scraping Reddit and trying to find

relevant things from like local llama

because I go in there I'm like certain

things I are way above my head and other

things are relevant so but it yeah it's

about what is relevant within within the

industry like I think the like mid

Journeys the dollies the everything that

is really interesting really cool I

can't put mental energy towards it it's

not in my in my chord directive and it's

not really a distraction but it's kind

of a distraction because there's just

there's Infinity rabbit holes you can

fall down so trying to maintain some

degree of focus while still you know

just pursue something you love add a

little bit of exploration on top of that

but you got to be somewhat focused

because every day there's new

developments yeah I think I don't know

was testing out I think it was Blue Sky

I need to stop I need to be like put

blinders on like a horse um I think it

was Blue Sky I was like I I don't care

about your AI cartoons like I darly and

all of that like they're so amazing but

I don't I can't bring myself to care and

then I feel like I shouldn't poo poo

things like that but yeah I'm going to

pop in the show notes open interpreters

Discord and um GitHub but where should

people reach out to you where you you're

most accessible on Twitter I feel like

yep I and in the two you mentioned um

anyone who's looking to get exposure

whether it's like serious involvement or

just a little bit check out the the

GitHub um some good discussions there

it's a good way to get familiar with the

product um we do have an an X Community

where you can get a lot of tutorials and

guides in a more Consolidated form

rather than just like Twitter as a whole

um and the Discord um earlier today I

hosted a build with Me session where I

was doing just like some code

refactoring um few people joed in we

worked as a team we got a PR pushed up

so really cool the community that's

developed open interpreter has been

wonderful for bringing people together

and just kind of like here's the core

Mission everyone wants to accomplish it

build each other up no drama it it's

been very good that's awesome thank you

so much Mike I'm really happy to have

had this time with you this was awesome

thanks for having me thank you that

wraps up this week's episode of

unsupervised learning I'm your host

Renee and I've had a great time chatting

with you as always links to everything

we discussed will be in the show notes

make sure you reach out to our guests

questions or feedback reach out to pod

unsupervised learning. until then leave

a like follow or rating on Spotify Apple

podcast or YouTube and until next week

stay curious