'Should We Be Concerned?': Josh Hawley Asks OpenAI Head About AI's Effect On Elections

Sen. Josh Hawley (R-MO) questions OpenAI CEO Sam Altman in today's Senate Judiciary Committee hearing.
Fuel your success with Forbes. Gain unlimited access to premium journalism, including breaking news, groundbreaking in-depth reported stories, daily digests and more. Plus, members get a front-row seat at members-only events with leading thinkers and doers, access to premium video that can help you get ahead, an ad-light experience, early access to select products including NFT drops and more:
account.forbes.com/membership...
Stay Connected
Forbes on Facebook: forbes
Forbes Video on Twitter: / forbes
Forbes Video on Instagram: / forbes
More From Forbes: forbes.com

[@portcybertryx222]

This is exactly why we need more young senators who can actually go out there do their research and then guide policy. My respect to Josh Harley. He may not have the same political viewpoint as I do but I respect at least having the courtesy to do background research and understand the arguments unlike other hearings like the Facebook one where everyone was clueless

[@CholeCoper]

Exatly!

[@James-cd7uu]

Keep it up Mr Hawley. Thanks

[@ruthegan8524]

My hero!

[@andysuter2809]

Enough with AI, I want to thank my man Josh for saving my manhood and instilling traditional masculine virtues.

[@jiiig8667]

People need to know if they are looking at AI generated content.. Full disclosure.

[@Douchebagus]

Will never happen. AI technology is in everyones' hands now, and anyone can generate it and hide that it was in fact generated. No legislation can ever change that or ever had a chance to change that.

[@ordinary1069]

Yup. There like needs to be a digital marker in AI art, or a disclosure on AI generated articles.

[@JaymeeKay]

I wish more people would see this.

[@GeoMeridium]

I have a lot of respect for Sam Altman. I might not agree with all of his choices or views, but he's very humble, transparent, and clearly concerned about how his company affects the world.

[@Billy420-69]

His LLM only parrots democrat propaganda.

[@carlosmspk]

Definitely. I can totally see how that could just be a business move, i.e., by coming off that way, he's more likely to gain the trust of future clients, however it doesn't seem like a business move at all. He's either very good at fooling us, or a rare case of a good successful person

[@hansmarheim7620]

I am impressed by the members of this committee. And their questions. Its almost like these politicians actually have a brain. I didn't see that coming.

[@JustDisc]

It’s funny how we all talk about properly preparing for AI, when human history shows is that we don’t learn/make changes until something happens.

[@paulettecole9482]

Problem is you can't always trust people to do the right things! That's whats scary, we definitely can't trust our Government!

[@barbarasizemore5961]

Not anymore!

[@AnnieNelson-wo6bm]

Never could they are still being switched and controlled now who is doing that from where. The evil things we hear coming out of the mouths of these people are not of thier own accords. No one can trust anyone they don't know

[@lapetusX]

these names...

[@RobotsPatriots]

What steps are taking to do something about instead of commenting on KZbin?

[@TeeTiz]

At least the Government can be held accountable (to an extent). You CERTAINLY cannot trust Corporations (private or otherwise), as their memorandum is profit less ethical and humane practices…

[@finnaplow]

im happy to hear a senator so well versed in what the implications of this might be. gave me hope though it may not matter

[@romandevivo1163]

Josh, AI aside, as you are well aware insurrections also have a tremendous impact on elections and the democratic process in general.

[@mattturner5429]

Well, then it's a good thing we've never had any insurrections. I'm sure Josh will give your comment a thumbs up, though.

[@thejustinjustin1233]

I respect Sam for being honest about the dangers of AI and where it could lead. We are in uncharted territory and it’s going to advance faster and faster to the point where it’s indistinguishable from human behavior and nature.

[@nomicwave]

Maybe that's a good thing. People might actually stop believing anything and start thinking for themselves again.

[@JevoKitano]

AI is usually not designed to stop at human capabilities. It might be indistinguishable for a second and then it'll keep improving, millions of times faster than biological evolution.

[@DanFromIran]

​@@nomicwavewhat? Believe?!! Loool

[@stevejones4431]

No, it won't be indistinguishable from human behavior and nature. You're conflating algorithms with AGI. We're not even close.

[@JevoKitano]

@@stevejones4431 ChatGPT is neither an AGI nor an algorithm though.

[@ronmills8092]

People like this can talk all day, and not say anything. That's talent.

[@Tates684]

😂

[@OiVinn-eq1ml]

I'm surprised Forbes would have this video in only 480p

[@pavelsmith2267]

We should stop with simply saying, Missouri is a great place.

[@ginagina5452]

Oh look, it's Josh Haulass. Run Josh, run. 🏃🏃🏃🏃

[@dagbomb7745]

I enjoy how government screams about ai and how nobody understands it yet people who do understand, and have shown they have the interest of the American people with them (like, Snowden), are called criminals and treasonous.

[@ginamarie--theuniversalson8518]

Josh/Mr. Hawley … your voice sounds fantastic!!! ❤

[@mitchellpugh497]

The potential for good and evil is a reality even in our own creations but only if offer it free will. Definitely wise to apply some restrictions. 🙂

[@spaztor7723]

good and evil are just subjective illusions, no matter what you do someone gonna think something is bad and someone else will think that same thing is good, so even regulating it wont work cuz u just cant please everyone. because nature don't work like that. but as we continue to fight nature, the more freedoms we will lose as a human

[@ruthegan8524]

Regardless of our advanced times and its technology, we need above all people who act with integrity. Our founders said that this system (a bold and brave experiment for them) will work -if the people are moral.

[@Tates684]

@@ruthegan8524I agree

[@AnaRanja]

Yes

[@michaelo.1320]

That guy behind him lmao 😂

[@mikelee9886]

Destroy it all. At a minimum, isolate it completely

[@bettyreynolds5106]

We should have been asked for permission from every individual to gather our personal information.

[@abram730]

You were. The terms of service, you never read.

[@ruthnovena40]

We should be the person who began it quit.

[@victorledezma6652]

Wow these politicians actuales asked great questions

[@roobler8923]

soon there will be hearings on quantum computing too

[@zyfryth]

So, about the AD-s... Does that mean that, the AI will be able to determine that I despise all AD-s and simply never show them to me, for they are pointless? 😂😂😂😂

[@andresinchausti3841]

the guy in the background xD

[@forthehomies7043]

The hardest part is going to be getting the Ai to tell the truth. I don't care what it says to me as long as it is the truth. If your political views are swayed because of the truth, that's a good thing in my opinion.

[@DeplorableSon1776]

We are not tryin...but they are

[@akharuzzamanovi]

New new ideas, angles are flowing

[@bludclone]

Tobuscus really got his act together

[@vineninja5882]

1:30 "I wanna think about this in the context of elections" Got me smirking

[@GjaP_242]

Soon after the covid-19 pandemic was declared, the World Health Organization signalled that artificial intelligence (AI) could be an important technology to manage the crisis caused by the virus. While AI holds much promise it also raises serious questions concerning fairness, reliability, accountability, privacy, transparency, and safety. 1:31 [The BMJ]

[@garymesser]

I had a cool idea with Having GPT-4 trying to decode the zodiac killers letters that never got cracked...I think that would be a cool idea to try

[@dmitri2366]

Cool idea, but chatGPT's model is unable to come up with original idea's and is limited to the dataset it uses. Therefore it cannot come up with something new outside of the dataset it already has been fed. Also ChatGPT is a LLM model and IMO there are better models to try and solve this problem.

[@pavelsmith2267]

The Altman family is renowned for breeding horses. We are being taken for a ride.

[@FST2375]

Guys from senate, the only way to answer all these challenges and concerns is to regulate it by forcing Open source. Non-intuitive like counter-steering

[@marlenepowell6369]

Chip Roy and Josh Hawley for 2024.🙏🙏🙏🙏🙏🙏🙏🙏❤❤❤❤❤❤

[@danielgrantcoleman]

Bro. All I’m asking is who the hell is an undecided voter?

[@tianrunde7534]

Dangerous artificial intelligence!!

[@ruthegan8524]

Needs close supervsion by REAL HUMAN BEINGS WITH INTEGRITY.

[@andramalexh]

The only way to combat ai would be to give everyone there own transparent ai to detect if an ai is hazardous. One that is not connected to the internet cannot be manipulated. Just offers "I have detected this ai is using harmful algorithms I would be cautious"

[@RealPatelGroup]

Imagine a company has the unfair advantage to fine tune new laws with the government to make competition impossible, smart move Sam.

[@marklivers5907]

do you josh hawley read comments on here,cause i need to ask you some real questions, and i would love to hear your answers

[@lilbaz8073]

We've released our product. Now introduce legislation to stop competition.

[@ashrafk8575]

That guy behind Sam looks like he’s the real writer of ChatGPT…

[@standtallvets5386]

It's all about very sick garbage, and needs to be treated as such!!

[@sh1nyarm0r]

How can you regulate a programmer? No way.

[@TheMexicanObserver]

Isn't that the same guy of the video " squish that cat" that ceo looks the same

[@dntwantgglplus]

what different between a voter being manipulated by existing forces vs gpt-4?

[@TimParker-Chambers]

''Greetings and salutations: Welcome to the emergency line of the San Angeles Police Department; if you'd prefer an automated response, press one now...''

[@merakim1824]

I'm open to new advancements though with it comes some concerns just like anything. I tried it with writing a short book and had to get more specific with my prompts and edit with it with more requests. One thing I didn't favor is that when you ask if a certain word is a real word it confirms it is. Then you ask if it is really and only until about third or fourth time it agreed that word wasn't a real word. I said it was a neologism. And it finally confirmed it. Now if someone didn't know that, they would take it as a fact. This is concerning for others who may take the responses as facts, knowledge when it is not at times. I noticed humans still need to provide prompts, ask for specifics, discern, redirect, and fact check where a disclaimer is needed for many individuals.

[@dr.sherbaz5900]

Absolutely! It's a fantastic concept to create a fact sheet that summarizes the data used to train a language model. This way, we can take a step forward in effectively utilizing AI models in a professional capacity.

[@lilbaz8073]

How long would the sheet be though? It would take an ai to read it.

[@racistindian5316]

That will work only as long as it does not has access to internet, once you give it the access it will start learning from everywhere. Also without internet access no one really can control it. We still don't understand completely how it actually behaves.

[@lilbaz8073]

@@racistindian5316 bard is linked to the internet. As is bing as is autogpt.

[@dv6165]

Of course American politicians only care about the impact on their own game. Has anyone considered to ask about misalignment??

[@venkarri8534]

These two could be brothers. The Missouri brothers.

[@mikelee9886]

How about just destroying all of the AI hardware and software NOW

[@Askejm]

sam altman: omg ai is so dangerous!!! please put in regulations to enforce my monopoly, especially against open source projects that will beat me in the future. thanks!

[@goldenfishes3695]

upvote so your message gets read. The only way for the common people to win is to hold back on govtn regulations and let the companies fight it out. Only then will the common lay person have a fighting chance.

[@erikstekelenburg3020]

Guys like this, both men, need to keep going this. On track! Stand up people! It's a Masquerade of Fools. 😎

[@RobShuttleworth]

Need to see how AOC feels about it.

[@codylim1]

Finished watching 10:02

[@user-vf3rr4by8p]

Any software architect understands that ChatGtp's architecture is spaghetti and no human could possibly keep track. I think 1 solution would be to keep it small, collect results -> insert to the next system as a human after review - > collect .

[@Paige-bz7xd]

They need training on the law...

[@Rayblenn]

Think of that dystopian possibility where we could all be potentially unemployed and relaxed at home and the only ones working were our politicians and they were required by law, to feed us. We need that someday, where it would be harmful to be a politician.

[@duinay3]

Can AI tell how effective our congress is? 😂

[@Paige-bz7xd]

Concern in what way?

[@Tates684]

He sounds like wily wounka

[@cinematic_monkey]

Good luck regulating bad actors. What an ego on that guy.

[@begreatbeu]

Its amazing how far the world has come - Computer - Internet - AI. What will be next? AI arguing and debating in politics and courtrooms for their rights LOL!

[@williamnoam8534]

I was wondering why my nvidia stock went up 21%😂 8:20

[@newmanpc4253]

Generative AI models are trained on copyright-protected data - is that legal, is it sampling, scanning. How can i protect my art. How can you say no to your art be scanned by ai???

[@UboaSan]

you can opt out, but they made it complicated and you have to make an account and many other steps for them to even have a look. Also it's not really scanning or sampling, but that doesn't matter since the use of private content for a dataset should be illegal even if the output can be transformative, you can still make it do unethical things that can damage someones reputation or steal someones work if you use the right inputs. Like a wise person said ''Ask for permission not forgiveness''.

[@mohammadchoudhry5183]

Hawley lives in Northern VA

[@normalwhiteguy1555]

3:24 dude sleeping lol

[@DrJanpha]

How come other AI chiefs were not asked like Sundar Pichai (maybe I have missed it?

[@renemarin3310]

Imagine this models that predict wall street who is using this products. Human manipulation is happening. We are slow on catching what they do. They have years over us.

[@jeffscott8516]

When the inventor of Ai quits his position with google so he can be free to warn the decision makers that AI has potential to destroy this planet we all need to take notice.

[@WolfyJ]

If your talking about Geoffrey Hinton then while media portrays him as the godfather of AI he did not invent it.

[@joeslowmama2048]

pause it at 2:38 Hawley looks like Andrew Jackson LOL

[@vikingarab99]

.."Photoshop....this will be like that... but on steroids" - killed me 🤣

[@mr.4ut109]

😂😂😂😂😂

[@mountainmama4133]

They probably could have used it in the last election. Just me.

[@franc7188]

This only matters if you’re a person who can’t live without being connected. It’s a choice. Go spend some time outside and leave your tech behind. And remember that the robots are only scary if we let them be. They will only do what we tell them to do. When you do have to use it, make positive and beautiful choices. It will make the world more positive and beautiful. The robots will only do what we tell them. So let’s tell them a better story. Because we are currently experiencing the effects of an unbridled internet that is fueled by power and greed by manipulating our emotions to convert. We must regulate this world and give it laws And we need leaders who have made it to help it work for the better of humanity For the good of the people That’s the work to be done

[@ordinary1069]

We need more people like you who actually care who have this attitude.

[@franc7188]

@@ordinary1069 I am giving my life to this work. We the people are in need of a better way

[@tonyjones4582]

@@ordinary1069 wrong, more people like him (although it sounds nice and i wish it to be true) will only make it worse. He states "The robots will only do what we tell them" is false, hence why even the godfather of A.I is warning of the rapid pace of a advancement to AGI - where the robots will do what they want...

[@John-Is-My-Name]

It scares me a bit listening to stuff like this. Because it sounds like the government doesn't really understand the vastness and importance of this. They listen to the scientist explain for example a human extinction event and then aswer back "could this be used by bad actors to get monetary gain?". What does that matter if there is no humans left? they dont seem to really understand the biggest fears.

[@jdanorthwest]

Many people will reject the truth from humans they don't like and believe a lie from those they do like. Does anyone really believe AI will have much of an impact on that group?

[@eulogy4anange1]

Love Hawley 1984

[@otisdriftwood8469]

Of course hackers will get into these ai models, and all hell will break loose. Just watch.

[@antonychrist3539]

What people do not realize is that AI has been there for a while already in our day today life. Its just that it has come into spotlight now.

[@robertseaborne5758]

Artificial intelligence is one advantageous thing, Gaian intelligence is another primary and thereby far more advantageous thing; get it?

[@Homeoffersunited]

In context of elections!! Hahahahah that’s all they can think about

[@richardrogers156]

Tear down the modern day tower of babel-confusion.Its that simple 🤔⚒️💪😇🙏🙏🙏🙏🙏🙏🙏

[@Atombender]

Sam never minces worlds and never apologizes.

[@kenbajema]

AI is an arms race for machine super intelligence. The analogy might be that the only thing to stop a bad guy with a gun is a good guy with a gun. Countries have taken this a step further. We have Nuclear weapons because other superpowers do too. But the coming capabilities of machine General intelligence is inconceivably greater than our own. AGI is like giving every person a nuclear weapon. Each individual does not have the same goals and we can't apply our own reason and values to others who may have been raised in a very different culture. This should terrify us all.

[@icantfindausernamehe]

Transparency, ingredients... we need to steal your technology . PMSL

[@TiagoAlexandrec]

Lol ibm talking about transparency 😅😅😅😅😅

[@annichka8711]

No video, no text can be trusted anymore. Could be fake, could be real, or a mixture. This has some extremely bad implications.

[@mikasa_sucasa]

Senator Hawley's question isn't good. Here's an example to compare it to: say, we all lived in complete darkness, and the invention of the flashlight was on the verge. Senator Hawley's question implies that there is an issue in the existence of flashlights. Instead, this is a question on government itself, about who should be issued flashlights. And the answer to that is that we should probably all be trained in using flashlights and that flashlights should accommodate for various abilities, so that there is equitable access for everyone.

[@DonC11669]

Less AI less room for fraud.

[@Less1leg2]

Interesting comment. When do you know if you are speaking to a Department person or an AI replication entity. If you can replicate a Rappers voice to such an extent that there is no discernable difference between "real to artificial". Then, how do you as a citizen know if you are calling into your government agency office you are speaking to a real telephone person or a computer generated telephone AI sound receiving device. If you are let's say a Clerk within Government, this will spell your doom. Why have Office Clerks when all you need is a single place telephone reception processing office.

[@armywashere]

In the realm of laws, where shadows intertwine, There lies a truth that we cannot define. The tools they possess, creations of distant lands, A haunting reminder of exploited hands. For those who enforce the code, with authority they stride, Carrying items woven in the threads of divide. From factories unknown, where dreams are displaced, Their labor undervalued, their existence erased. The objects they wield, a symbol of their might, Yet crafted by the oppressed, hidden from sight. The fruits of exploitation, veiled in the night, A truth obscured, but still shining bright. In this dance of complexities, let us be aware, Of the systems we partake in, with utmost care. For the items we possess, a reflection profound, Of the interconnectedness that knows no bound. So let us question and challenge, as the shadows reveal, The chains that bind, the wounds that never heal. For a world that transcends the cycles of pain, Where fairness and compassion forever reign. In the realm of laws, let empathy guide, As we seek a future where justice does not hide. Where the echoes of exploitation fade away, And a brighter dawn illuminates our way.

[@Barbatian]

More than AI I’ve always been scared of humans. So far we are the greatest threat to us and everything around us. I wished we explored more ourselves than everything around

[@jeffscott8516]

If you want to put all our apples in a algorithm then you will have no control if they logically cross us out.

[@Sandsplans]

Congress be like: how's AI gonna affect me getting elected

[@politicalmoderate190]

The problem is that the folks in the AI industry are fighting for funding, clout, and the authority to develop in any direction they want. To do so, they have to find the Senators that are the easiest to scare with science stuff and keep them terrified. The fact is that the best AI that's been produced TO DATE with language learning software can be found out as an AI relatively quickly. AI's, in general, especially the ones today, are remarkably stupid compared to the AIs that people THINK are being developed. Nobody wants to hear "eh, it's not as bad as all that," so it's constantly a glass half empty scenario.

[@insipylipy]

Well said.

[@lavs8696]

It's important to have these discussions now than after it's too late, so no, these aren't just scare tactics. No one is saying we have Terminator level AI systems right now, as merely things like AI generated pictures are already a massive cause for concern. And it is very much an exponential growth curve when it comes to AI development, so we have to be ahead of the curve before it starts accelerating into something that could greatly destabilize our systems. AI can be an incredibly useful and powerful tool, but it can also be an incredibly dangerous one, and it is up to the government to create guidelines to limit the possible downsides. These meetings are very necessary, and your attempt to downplay it is naive.

[@dmitri2366]

Disagree, I am a software engineer and I have been following AI for a while before the hype. It's actually quite scary how fast this field has advanced in the last 10 years. It's already pretty clear that this will become more and more advanced. I am not saying that AI will take over the world. But whoever will have control over the most used platforms will have control over the information we receive, our educational systems and much, much more. The dangerous part about these systems is that they are a like black-box. You give it some input and output is rolled out. However, you receive little to no information about the process itself and the dataset that is used to train it(unless its open source, fairly uncommon rn). It is very easy to exert control over people this way. As most of us won't be questioning what the models spit out as output once this becomes a standard.

[@politicalmoderate190]

@Dmitri every single modern AI, ChatGPT included, is a regression engine that makes educated guesses. At its core, it's y=mx+b and even then it's 85-90% accurate based on the info fed in.

[@politicalmoderate190]

@Dmitri and you're talking to someone with a Masters in Data Analytics with a focus in Machine Learning and Artificial intelligence. Any model that has over 95% accuracy is red flagged for overfitting data and is usually discarded. Additionally, most AI construction (most, not all, because there are several that are simply elaborate if-then engines) is based on the Neural Net pathway model which builds up layers upon layers of decision trees and filters based on thousands of data points (on the low end). These trees and filters, if its built correctly, can also be set up so that getting something wrong refines the model to include more samples that were false positives or negatives, but again, you run the risk of overfitting the data. Licensing means the government determines which sets of developers get to move AI in what direction: right now, it's pretty aimless, but for a new technology, that's fine. Letting the government decide who researches will likely restrict it to suitably large corporations or the DoD, and I can't help but think that'll pigeonhole most of the applications.

[@webdev1876]

AI is careless where you were born

[@tjkool]

Everyone needs to chill with all of their hysteria over OpenAI.

[@gregorymabrey7508]

It sounds like these experts are saying we need to decide whether or not human beings future masters will be kind or mean. We humans are programming machines to enslave future humans 😂😢

[@NO-GAMES]

Can you get your head around AI replacing everyone in government and eventually killing off humanity?

[@bob733333]

Ya, just blame it on AI and cover up for the nazis.

[@josflorida5346]

Oh shit ai heard you ^^^^ 😆

[@normalwhiteguy1555]

look like 2 wax models talking pretty much AI