00:00AILA is an AI agent, AILA.
00:04What is AILA AI agent?
00:08What is AILA AI agent?
00:12We have also created AILA Artificially Intelligent Laboratory Assistant,
00:17which is a laboratory copilot.
00:19If you use the chat GPT or language models,
00:23you can type it.
00:25If you ask questions, you can answer it.
00:29In AILA, you can type it in the English language.
00:33If you have experiments in real world,
00:36you can analyze the results of the results.
00:42What is the result of this problem?
00:46What is the problem in our country?
00:50One of the problems is skilled human resources.
00:56There is no skilled human resources.
00:57There is no skilled human resources for experimental facilities.
01:02So, AILA can be used to upskilling,
01:04you can train humans,
01:06and any experiments can be done automatically,
01:09so that humans can be focused on scientific things.
01:14And any experiments will do AILA.
01:17What is the problem in human resources?
01:18What is the problem in human resources?
01:20We have a proof of concept called Atomic Force Microscopy.
01:22What is the problem in human resources?
01:24We have a proof of concept called Atomic Force Microscopy.
01:28But as many materials,
01:31if you think about energy,
01:33or medical,
01:35or agriculture,
01:36whatever fields,
01:37you can use materials,
01:38you can use AILA.
01:40And the strength of materials,
01:42the fracture of materials,
01:44the problem,
01:45the problem,
01:46the problem,
01:47the problem,
01:48or the problem,
01:49the problem.
01:50So,
01:51what will the result
01:52be the problem?
01:53How many percent will it be?
01:55This is a very important question.
01:57Because when humans experimented,
01:59there is a problem.
02:00There is a problem.
02:01Or,
02:03if we use language models,
02:06there is a problem.
02:07There is a problem.
02:08So,
02:09we have designed these 100 experiments,
02:11and evaluated,
02:13how much of the failure rate is.
02:15And,
02:16in GPT,
02:17we have found that,
02:18we can successfully do almost 80% experiments,
02:20but,
02:22we have also 20% failures.
02:23But,
02:24the important thing is that,
02:25when a failure happens,
02:26we will tell you that,
02:28the experiment has failed,
02:30and it has failed.
02:32So,
02:33we will know what happened,
02:34when we experimented.
02:36So,
02:37the result of the experiment,
02:39the result of the experiment,
02:40is it the final amount,
02:41or is it the human interest?
02:43No,
02:44when the experiments finally get,
02:46the results of the research,
02:47they can rely on it.
02:48They can rely on it.
02:49They can rely on it,
02:50because,
02:51in the real world,
02:52the experiment has been performed,
02:53and the experiment has been made.
02:55So,
02:56after that,
02:57if you have to analyze it,
02:58then,
02:59the human analysis can be done.
03:00Or,
03:01if you use this research,
03:02then,
03:03then,
03:04it can also be done.
03:05So,
03:06there have been a lot of efforts,
03:07to develop language models.
03:08So,
03:09you know,
03:10Meta,
03:11Google,
03:12so,
03:13so,
03:14there have been a lot of efforts,
03:15to develop language models.
03:16So,
03:17you know,
03:18meta-language model,
03:19and
03:20Google-language model.
03:21So,
03:22this language model,
03:23there are more efforts,
03:24to use online search,
03:26or to retrieve information,
03:28and to get information.
03:29We have collected this,
03:30in a real world,
03:31experimental facilities,
03:32and this is one of the first,
03:34of its kind of efforts in the world.
03:36To connect it,
03:37in a simple English language,
03:40in English language,
03:41and they will give you an experiment,
03:42and they will give you an experiment.
03:43This is one of the first,
03:44in the world.
03:45Sir,
03:46if this is a success,
03:47they will give you an experiment,
03:48then,
03:49what will the effect of people,
03:51in the world,
03:52and other industries,
03:54in the world?
03:55So,
03:56as many industries,
03:57like MSMEs,
03:58or small labs,
03:59there are no resources to do that.
04:02So,
04:03a lab,
04:04using such facilities,
04:06a small R&D facility,
04:08or up-stilling,
04:09so,
04:10our mission is,
04:11democratizing scientific research,
04:13democratizing experiments.
04:14So,
04:15with very less cost,
04:16and with high throughput,
04:17and with high throughput,
04:18you can use these experimental facilities,
04:20in research.
04:21So,
04:22in research,
04:23there are a few months,
04:24and months,
04:25and months,
04:26so,
04:27if we research them,
04:28how much?
04:29So,
04:30the first advantage,
04:31is that,
04:32it is 24-7,
04:33so,
04:34it doesn't need to break,
04:35or leave,
04:37so,
04:38basically,
04:39it will work 24 hours,
04:41continuously.
04:42IELA,
04:43how does IELA work work?
04:44Tell us a demo.
04:45So,
04:46IELA,
04:47we have made a software,
04:48so,
04:49we have made a software,
04:50we have,
04:51here,
04:52there,
04:53there,
04:54there,
04:55there,
04:56there,
04:57and the materials,
04:58what,
05:10write a system,
05:11so,
05:12you might be able to do this if you do this like,
05:17If you want to study the properties of any metal, you can study the properties of battery, energy storage, devices, different properties of battery storage,
05:26then you can do it with the nano level.
05:28Okay, let's do a little demo.
05:30For the demo, if you want to see here, we have a prompt that we have to do P-Gain 50, I-Gain 20 and D-Gain 0.
05:37So if you want to see here, P-Gain 50, I-Gain 50, I-Gain 50, I-Gain 50 and D-Gain 50.
05:42So I will add a prompt address and I have found a GPT-4O model.
05:46You can choose different models.
05:48So I have a GPT-4O model and I will send it to me data.
05:52So you can see that I-L-A called.
05:55I-L-A talks from different agents.
05:58After talking, ultimately, the code is written.
06:01And you can see that the final change is 50, 20 and D-Gain.
06:04And you can see that the answer is 50, 0, 20.
06:09So you can see that the answer is 50, 0, 20.
06:13So this is a change.
06:15So simply, we have to write a prompt.
06:19So the prompt is converted to the experiment.
06:21If you want to take an image and reference it,
06:25then you have to write the same image and measure the reference.
06:28So this will be done.
06:29For what you have to do, you have to take a lot of steps.
06:32You have to take a lot of steps to the experiment.
06:33So you have to take an image and take it.
06:35Then you can take it in another system.
06:37Or you have to open another system.
06:38You have to open a system and analyze it.
06:40You have to do everything in the system.
06:41You have to do everything in the system.
06:42So this is a process.
06:44And how can it use it?
06:47This is a use of academic post microscope.
06:52We have used to automate it.
06:54We have used to automate it.
06:55If there is any instrument that has a proper manual available,
06:58then we can connect the instrument with it.
07:01So that's why we need a proper manual.
07:03We will teach it how to use the instrument.
07:07So we can integrate it.
07:09I mean, we have to talk about some things.
07:11Like the environment.
07:12We have to talk about the producer.
07:14So how will the producer give it?
07:17You have to do production.
07:20You have to measure the air quality index in real time.
07:25So you need an air quality index detector.
07:29So if you have a detector,
07:31if there is a proper manual,
07:33then we can connect it with it.
07:35You have to take the manual.
07:36You have to teach it.
07:37Then it will connect it with the detector.
07:38Then you can measure it with 24 hours.
07:40You can measure it with pollution.
07:42If the pollution level is going up,
07:44then there will be alerts that the pollution level is going up.
07:47So we can do everything.
07:49we will start to do programming.
07:51Yes, we can do programming itself.
07:52We can learn it.
07:53Then we will learn it,
07:54there will be different process of how we teach it.
07:57So let's use the manual.
07:58It will take the particular process of manual.
Comments