Chapter 11

Chapter 11

Chapter 11

Matt thought about Griff's words for the next few hours. He didn't know what to do. The risk was fairly high, but the reward was also great.

If he reached out to get the specialized researcher AI, he would have an AI to use, but in the process he risked letting important information slip. It was as if he was shouting that he had a unique mana producing Talent from the rooftops, and that was dangerous.

Mana was precious on higher Tiered worlds, where the general population didnt just live by the safe areas along the coast. The runic formations that stopped rifts from forming in cities cost a lot of mana to run. The cost was only partially offset by the higher Tiered cultivators and rifts producing more mana and mana stones.

As Matt contemplated the risks and benefits of the AI, he leaned more towards doing what Griff suggested. It was dangerous, sure, but he didnt think even massive companies would dare risk the Empires wrath. His status garnered from being on The Path would protect him.

Research first. Matt looked up what he could about the company TrueMind Corp. What he found lined up with what he expected. Would a company that large let negative publicity ruin their image? He didnt think so.

What he did discover was that their main source of revenue was producing Tier 30 and above smart runic formations. The selling of personal AIs was a side project subsidized by The Empire. They stood to benefit from the advantages the AI would give their younger generation. Shadi, the founder of the company, was reportedly altruistic in her old age. That was after she had crushed the competition, twenty thousand years ago.

Now that TrueMind was alone on top of their target market, the founder reportedly had given up on advancement of her cultivation. Instead she had been focused on a personal project and rarely seen. Matt didn't trust that at all, but he doubted she was up to anything too heinous. Besides, producing a few mana a second couldn't be that rare at higher Tiers, it was just uncommon at low Tiers. Any benefit the founder would gain from taking advantage of him wouldnt outweigh the consequences.

It was still a risk, but the reward of a personal AI was worth it. The AI would always be useful, unlike the other expensive items Matt needed to progress. An enchanted sword would grow obsolete after Matt reached Tier 4 or 5, and a spatial bag had the same problem. The bag would last longer than the sword, but still be replaced at some point. Most Cultivators used them even after Tier 15, because the lower-ranked spatial rings had a limited storage area. However, it was said that eventually the bags were completely replaced with rings after a point.

He couldn't always plan for Tiers ahead of his own, but it felt like a wasted opportunity to not get the only item that would grow with him.

Making his decision, Matt called the local offices of TrueMind. When the receptionist answered, he simply asked to speak with a manager, and was transferred.

When the person picked up Matt asked, Are you someone who can make decisions about unique Talents and AIs?

Nope, this is the complaints office. Let me transfer you. Matt facepalmed. He had immediately asked for a manager, so it made sense that the receptionist had sent him to the complaints department.

It was a much longer wait this time.

Hello, this is Janet with the development team. The voice on the other side sounded only half attentive.

Development team? What were they developing on this low Tiered world? Matt wasn't sure, but gave her his best pitch.

I was informed TrueMind was interested in unique Talents. I happen to have one that produces a lot of mana under the right circumstances. But that also makes me unable to use the standard AIs.

Janet sounded more interested when Matt said that. Really now. How much is a lot? If it's enough we would be willing to hire you.

He had already decided to not reveal his true powers, but he also realized he had to show a little of his abilities if he wanted their attention.

I am not able to work for your company as I'm on The Path of Ascension.'' He wanted to get that out as soon as possible, so they wouldn't get thoughts of crossing The Empire to get his Talent after what he said next.

And I can produce around humm... Lets say five mana a second. No reserves.

All he heard was a chair creak, it was a long moment before Janet spoke up.

Ok. ok, that's interesting. So, some kind of high mana regeneration but lacking in capacity. That is unique. I assume that the trend will continue, or you would have just waited. I'm assuming you want the researcher AI, right?

Matt swallowed hard; those assumptions were too accurate for his liking.

Yes, that's exactly what I want.

For the next few minutes all he could hear was rapid tapping at a screen.

If you want to purchase a researcher's AI, I can do that. It will cost fourteen Tier 6 mana stones.

That crushed his happiness, that was three and a half billion credits. He didn't have a hope of getting the AI unless he could work out a deal.

I was informed that a deal could be worked out? Matt asked hopefully.

It could be, but it comes with some restrictions and conditions. Are you interested? Most would rather wait.

Matt didn't like the sound of that, but this could help him grow faster, and that was priceless.

What are they? He was happy his voice didnt waver.

So, I can get you the standard researcher AI and one software, but you have to come in to do scans and testing at every Tier until Tier 10. If you miss any because of any reason, you would owe us the missing Tiers after Tier 10.

The other option is something weve been working on for a while. It's the beta version of the next generation researcher AI. Its more versatile and built to be more malleable. The problem is its mana inefficient. Our job and another dozen worlds, is keeping the malleability while reducing the mana cost.

This sounded better to Matt, something new would be an advantage. Griff had said the newer models had a higher starting point. It was tempting.

The adaptability and increased computing power are expensive. Very expensive. If you hadnt said you generated five mana a second, I wouldnt have even offered it. The idling mana cost is a permanent one mana a second, and the active computing can multiply that easily.

So, if you exaggerated your mana abilities now's the time to say so.

Matt hesitated for a moment, thinking over all the scenarios he could imagine. Then with determination he said, No I didnt, that sounds like a useful AI for my situation. Whats the catch?

Janet chuckled. The normal catch is the absurd mana cost. We would give it all the software for combat and predictive modeling. Wed want you to push the AI, and heavy testing under combat conditions always brings out good data.

If you take this option, we would have you come in to scan until Tier 15, and wed send you a special pad with testing software. We would expect the AI to be tested on downtime. Also, Im obligated to inform you that we usually pay for scans and testing of cultivators AIs. The usual rate is one mana stone of their Tier plus one. So, youd be losing out on a load of income.

None of that bothered Matt, the AI could do the testing when he was asleep, which would also allow him to benchmark its growth. The income loss wasnt relevant, it was a lot of money now, but to the future him, who was a Tier 10, earning a Tier 11 mana stone wouldnt be impossible.

Matt made his decision. As long as the testing wont run into my own time I dont mind.

Really? That's good. I can send you the contract then. I gotta say I didn't think this would come across my desk today. It's good for our branch though.

As Matt read over the contract he idly asked Why is that? Do you get a bonus or something for finding a tester?

Eh kind of. If you prove even mildly useful well get a bonus sure, but this is an important project. It usually takes decades to get an AI good at something without incredibly specialized programming. And the programming limits the AI. The ones that learn naturally are far better than ones that get massive software programs.

Janet continued, with the passion she had for the job showing clearly. Its why it takes so long for a new generation of personal AIs to come out. We learn from the older AIs and can make the starting parameters even smaller and less detrimental to the AIs future growth. The older an AI gets, the more it can break the shackles of early programming. But its hard and can take centuries and an absurd amount of mana to do.

This design has been on our back burner for at least a hundred years. Really it's an interesting design philosophy...

Matt only listened with half an ear as Janet continued with how his data could help them in the smallest ways. As he reviewed the contract, he found it was as she had said. He would owe them a scan and test every Tier until 15, and if he missed one, he would have to make it up on the back end.

The other part about the AI having to run specialized testing was vague. It was written as a pre-established set of testing software that would take up no more than an hour a day under normal conditions. And he had a month to adjust to the AI before testing began.

After scanning the contract again for anything he might have missed, he digitally signed the document.

In his vision, he saw, Startup procedures initializing. Please stand by. Estimated time to finish, 14 minutes.

Matt took Aster to eat at the dining hall, and after setting out her food, ate his own. The new food was even more expensive than her kit food, as it was made of Tier 4 and 5 monsters. The expense drained his already tight budget, but he wasn't going to skimp on his companion.

The startup process finished, and a heads-up display overlaid his vision. New host detected. Start scanning? Yes. No.

Matt selected yes and felt the pull on his mana grow stronger.

This AI is absurd.

It was pulling almost three mana a second, any normal mage at this Tier would be empty of mana in just a few minutes. It did explain why they had been working on this model for so long. It was nearly unusable at lower Tiers, and at the higher Tiers no one would want to get rid of their older, more established AI for this glutton.

After a moment, the mana draw slowed down to a trickle, and Matt now had an outline of his body covering his view of the dining hall. When he looked closer his body was tinted green. Focusing on any part gave a breakdown of the limb and its current status.

Under that was a display for his mana.

Mana: 0.9/10

While the mana display wasn't very useful for Matt, the body scan was. It would be nice to be able to see any damage done to him.

Matt looked to Aster, who was cleaning herself, and the AI responded to his intention. Bond detected. Start scanning? Yes. No.

He went through the process, and a small outline of Aster appeared next to his own outline with the same breakdown. It hurt to realize that his fox had 900 mana at Tier 2, while he had a whopping 10.

After they finished their breakfast, Matt headed to the training room and set the room to the goblin and hobgoblin configuration of the Tier 1 rift. He made sure to bump up the difficulty, but only to peak Tier 2 strength. This was more about testing the AI than his own skills.

Once he activated [Cracked Phantom Armor], the AI prompted skill use detected. Start analyzing and optimization? Yes. No.

Matt smiled, this was exactly what he was hoping for. All AIs could optimize mana usage, but it was usually only seen at Tier 5 and higher. Hitting yes, he waited eagerly.

[Cracked Phantom Armor] detected, analyzing properties and mana usage. Processing. Processing. Processing. Analyzing complete. Optimizing. Failure. Please provide more usage data and information on optimizing skills.

Decidedly less thrilled, Matt realized he shouldn't have gotten his hopes up. It was a long shot to expect a brand-new AI with the mostly barebones software to do what took most AI years to accomplish.

He started the combat predictive mode, then had the AI start the training room scenario.

Immediately he stopped the training room.

The AI had shown every possible move the training aid could do. The predictions ranged from the general attacks to more obscure combinations the goblins would never do. It was blinding and cluttered to the point of being useless.

One prediction had even shown a .000000001% chance of the training aid just shutting down. That prediction was useless even in practice.

Starting to tweak the settings was tedious but easy enough. He lowered the tolerance for shown actions to forty percent or higher only, while taking into account past actions for the monster type and the general body structure.

He repeated the process again. It was better with less clutter, but after a few exchanges he paused the training aids again. Matt then re-enabled the predictions that accounted for the weapon type.

The next fight was better, but he stopped the training after a few exchanges to raise the threshold of shown prediction to sixty percent, and increase the transparency for all predictions greatly.

After that, Matt just fought the training goblins and hobgoblin. He wasn't sure if the learning speed was normal, but the AI was soon predicting the attacks with far fewer possible predictions. Even better, the most likely path was nearly always correct after only a half hour.

He didn't quite trust the predictions, because in the end this was a simulation of the goblins, not the real thing. While it was nearly perfect, the real goblins could, and did make unexpected attacks at times.

The training continued for the next hour, with Matt tweaking settings as he went. He ended up changing the transparency to indicate the probability instead of showing numbers as it had before.

Having the AI rereview all the data it processed, he went and purchased three new 200 mana rechargeable mana stones. Tun had suggested it so he could swap the power source for his new blade, and Matt agreed. He didn't want to get screwed because of the wait time for his mana to unaspect.

When he had asked about letting it use his own aspected mana, Tun had said that would double the price because of the added runes, so he just dropped the idea.

Tun said the weapon would be done in four more days, but that left Matt with not much to do for the next week, so he trained and exercised. Any down time was mostly spent stress testing the AI.

The pad TrueMind had sent had far more testing scenarios then what was mandatory. There was monster vs monster combat, monster vs cultivator combat, and even cultivator vs cultivator combat.

Most scenarios were just small snippets that the AI was supposed to reconstruct the fights off of. Some contained a small area left over from a fight, and some a vast, destroyed landscape with hundreds of corpses. A more interesting scenario he found was when the AI was put into a POV of a fighter, and had to predict the opponent's movements. The AI was awful at it, but Matt was interested in watching the fights play out and making his own predictions.

At first the AI had the same predictive problems as before, but as he let it watch the early fights, it slowly got better at predicting the flow of combat. It was still useless when it came to skills or Talents, but it got better after he found an information repository on common Tier 8 skill shards on the EmpireNet.

The AI could predict low Tier skill usage somewhat, but only after seeing the skill used at least once. So, surprise attacks and a clever first usage still tripped it up.

It was only at a five percent predictive ability with any fight that involved skills, and only two percent for combat over Tier 5. Still, Matt felt progress of his own watching the fights. Seeing the flow of combat in higher Tiers gave him ideas and possibilities for his own fighting style.

He also found an AI downloadable class on veils, and purchased it for 10k credits. It was incredible. The class had information packets that allowed his AI to simulate the feeling and pattern of what a veil should be.

With the AIs help, Matt was able to get a rudimentary veil that would dampen any spiritual sense trying to probe him in only a few hours. It wasn't perfect, as he couldn't hold the essence at the edge of his spirit as well as the AI projected, but he was happy with the progress.

It was a strain on his concentration, but would help hide his mana regeneration, so he set the AI to ping him any time he let the veil slip. The improvement was slow but steady. He and the AI predicted that within a few weeks, he would be able to hold a weak veil constantly. Then, it would just be a matter of strengthening it over time as his spirit grew stronger.

His new weapon was finished two days before the ban on his delving was lifted. Matt was glad he had gone with the more expensive version of the weapon once he got it. He and Tun had done a cutting test, and even without the enchantments activated, it was stronger and cut better than his first sword.

When the enchantment was active, it cut through a log like a dream. It was odd to use the enchantments on the blade, it felt like his spirit was trying to lift heavy weights while being covered with a wet blanket. It was a distracting sensation to get accustomed to.

It was mana efficient, being a low Tier weapon, but the more stress the weapon was under meant a faster mana drain. Still, Matt went and bought another three rechargeable mana stones. He just had to willthe runes active with his spirit, and they would pull from the mana stone in the pommel. Having extra was a nice backup.

With constant use, he could expect the two hundred mana to last an entire rift, but if he did run out of mana, he had extra mana stones to swap out. Using a rift mana stone was a waste of credits.

There was even a source of income with his mana if he was careful. Out of the PlayPen, people could sell mana to fuel the cities. The price was pretty low at fifty credits for two hundred mana, but it would keep him from starving.

For a moment, he wished he had had a rechargeable mana stone and device to transfer the mana when he was at Bennys. He could have made the ten thousand credits in no time.

The lack of information had greatly hampered his planning, though Matt figured he wouldnt have been at Bennys to meet Dena and Eric and join The Path if he had sold his mana. So maybe it was better he didnt know of the ability.

It's weird to think ten thousand credits was a far goal. Now, I can make it in a single rift if I get a little lucky. And with the Tier 3 rifts I'll make that with every mana stone. Yet I still feel poor. Things are more expensive, but Ive come a long way.

I can't believe I once thought that Benny charging 400 credits a week was a lot. Most of them were Tier 3. The place truly was a dump that was just in a good location.

His gear was starting to get a little small, so he took his time off to buy new clothes for combat and casual wear. The combat clothes he bought were simple, just sturdy cloth, but they fit better and wouldnt tear apart if he moved too fast.

***

Four days later Matt stood in front of the tear to the Tier 3 rift, and with Aster on his heels, he stepped through.