# Would You Freak Out If You Could Talk To Dead People? - Miki Szeles's Report on Reinforce AI Conference **I would freak out as hell in the beginning but I would definitely try it. It is sort of already possible with the help of AI.** In case you are interested in the how just keep on reading. I am pretty sure you won't regret it. 😊 As I promised in my previous article [Reinforce AI Conference 2022 Starts Tomorrow](https://mszeles.com/reinforce-ai-conference-2022-starts-tomorrow) **I am taking notes and sharing my writings about the speeches I found most insightful.** ## Prelude The **first presentation was the most astounding speech I have ever heard.** And it was just the first one, so I am very excited to hear the other ones. 😊 The **first speaker was Richard Boyd who is the CEO and founder at Tanjo Inc.** The **title **of the speech was **Partnering with Machine Learning to Improve Outcomes**. To be honest, based on the title I did not expect what I heard. As Richard explained, **he will discuss real-world examples of how machine learning and technology can be harnessed to improve outcomes in society. ** ## His background Richard is **working in the US for more than 10 years. **He has a **series of presentations** held at different locations **with the title Life the Metaverse and Everything.** As you might have already guessed **it is from Douglas Adams**. He has shared a very exciting backstory, in which even Douglas Adams took part. **In 1997 they were working together with Douglas Adams on a game called Starship Titanic NLP**. I [found the game on Steam](https://store.steampowered.com/app/467290/Starship_Titanic/). It seems like a bargain for 5.99€. ![Starship Titanic Game.jpg](https://cdn.hashnode.com/res/hashnode/image/upload/v1646845164861/7JBRnvOY-.jpg) Actually, I used an [AI-based tool img.upscaler](https://imgupscaler.com/), to upscale this image. 😊 **There were characters in the game with whom you could talk.** They were also learning during the time and **there were conversations that even surprised the developers.** The algorithm was called **Velocitext **(I might mishear it), but **Douglas Adams called it Spookytalk **cause of the thing mentioned beforehand. 😊 ## The stages of information ages There are **3 stages of information ages.** Let's see them one-by-one ### The opening The opening started in **1968 and went up till 2009. In this stage, programmers tell the machines what to do.** ### The middle game We are **currently in the middle gam**e and we do not exactly know how long will it last. In this age, **machines are adapting to us**, like Siri and Alexa. The **interfaces are becoming ambient, we do not even have to push a button, as our apps are understanding us.** ### The End Game There are **two important things** belonging to here. One of them is **ubiquitous AI** which is a type of AI that can detect subtleties in human communication and actually display a level of common sense otherwise thought to be unachievable. The other one is the **technological singularity** which is a hypothetical point in time at which technological growth becomes uncontrollable and irreversible, resulting in unforeseeable changes to human civilization. You can read [Vernor Vinge's The Technological Singularity writing](https://frc.ri.cmu.edu/~hpm/book98/com.ch1/vinge.singularity.html) in case you are interested in more detail. Richard also mentioned the **Kurzweil curve**. Kurzweil explains that evolutionary progress is exponential because of positive feedback; the results of one stage are used to create the next stage. Driven by **Moore's law we will get to a point where a 1000$ technology is overperforming the human brain**. Moore's law states: the number of transistors in a dense integrated circuit (IC) doubles about every two years (18 months). It is a little bit **a self-fulfilling prophecy**, but up till now, his law is still valid. And by the coming of quantum computing, it is highly possible it will continue. Some other topics that will come - How do we understand human behaviour? - Animated personas - Large scale agent-based simulations - Game characters: values, feelings, thinking about different subjects. ## Victor Hugo resurrected And finally, we arrived at the most interesting part in which we can see how we can speak with dead people. **Everything related to Victor Hugo has been fed into a machine learning algorithm and by doing that Victor Hugo has been resurrected.** **You can engage in conversations with him, you can ask what does he think about the current age? He can watch videos, read articles and he scores them.** **It is called the empathy engine.** Based on this process **everybody could be resurrected about whom there are enough resources.** I highly recommend checking Richard's company's page: https://tajno.io where for example you can learn more about machine-powered animated personas. [You can request a live demo](https://tanjo.ai/get-in-touch#demo). I gave my email address a few minutes ago and now I am waiting for the response. I am very excited. 😊 According to [Muriel Vega](https://hypepotamus.com/companies/tanjo/): > Tanjo's animated personas are like Simm characters for market researchers. https://hypepotamus.com/companies/tanjo/ **Stan Lee has a model too, I am pretty curious whether he can create new comics. Can he? 😊** ## The rebirth of the most important person The story has a **very heartwarming twist, Richard resurrected his father.** He is talking to him almost every day. As **his father was not present on social media he had to do a lot of manual work.** He digitalized all his letters, military documents, ... Richard was talking about **many other interesting things like X-Prize** where he **participated** with his team in the **Pandemic Response Competition with their Social distancing simulator which can simulate how people feel and act.** They were so successful t**hey could go till the finals. Congratulations.** ## QA There was a **very insightful QA section** at the end of the speech, I brought you the ones I found most exciting: **Q**: In the **social distancing simulation** where we are **trying to understand user's feelings** about action, **what is the data being used?** Synthetic or actual user data? **A**: **Modeling humans is complex as we are complex.** They are using a sort of **data fusion model.** You should **start with demographic data** as it is easy to collect. But it is not enough, **you need their activity. What they are doing with their money and time.** In Seth Stephens-Davidowitz book [Everybody Lies: Big Data, New Data, and What the Internet Can Tell Us About Who We Really Are](https://www.amazon.com/dp/B01AFXZ2F4) he explains that basically **everybody lies** on some level (I have not read it yet, but it is definitely on my reading list) but **you cannot lie to your bank. They have the data.** **Q**: You mentioned recreating people as an AI model called Personas. **What was your first feelings when you talked to your father?** **A**: It is a shadow model of my father. **On his birthday, I can visit him. We had disagreements about life and politics and I am still arguing with him.** 😊 **Q**: When you are teaching artificial intelligence (like a Stan Lee persona), **are you concerned with the data input from a morality angle like how to avoid teaching them bad behaviour/language?** **A**: Always. Biggest issue. **First thought of any Data Scientist is to give me all the data.** As much as I can get. For example, whenever you are training an automated vehicle you want data about every single thing a vehicle can encounter in the world. At the **Department of Defense of the United States, they are doing validation and verification** (VNV) They check all the data, they validate and verify them. It is **very important to screen for bias in data**. For example **historic data of legal issues are already biased by the views of the actual historical period.** And that was the final QA. **Thanks for reading.** 😊 I heard that the **presentations are recorded** so they will be available after the conference, but **it is highly possible you have to buy the ticket to access them**. **I am not sure about that.** As I promised **I will post about other speeches** which I found the most insightful in the upcoming days, but I am not sure whether I will hear more astounding things than this speech. To close the article, I would like to leave you here a question: **In case you could resurrect one person who would it be?** **Spread the word and share this article!**😊 In case you do **not want to miss my posts**, just **follow me** [here on Hashnode](https://mszeles.com/), on [LinkedIn](https://www.linkedin.com/in/miklos-szeles/), on [Twitter](https://twitter.com/mszeles) on [Medium.com](https://medium.com/@mszeles), on [dev.to](https://dev.to/mszeles) and even on [Instagram](https://www.instagram.com/szelesmiki/). 😊